CV/FCN.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 用FCN来完成语义分割任务\n",
    "\n",
    "\n",
    "## 1. 什么是语义分割?\n",
    "\n",
    "语义分割指的是**像素级别**地识别图片,标注出图像中每个像素属于的对象类别.\n",
    "\n",
    "比如这张经典的图片!\n",
    "\n",
    "![example1](./images/fcn_example.png)\n",
    "\n",
    "我们除了识别出自行车和人之外,我们还将他们的边界也进行了描绘.\n",
    "\n",
    "所以**语义分割可以看做对每一个像素进行分类**\n",
    "\n",
    "\n",
    "## 2. FCN的简单介绍\n",
    "\n",
    "为什么要叫 FCN(Fully Convolutional Networks) 呢?\n",
    "\n",
    "因为在以前的任务中(2015年之前)使用CNN更多地是在卷积层之后会接上若干个全连接层, 将卷积层产生的特征图(feature map)映射成一个固定长度的特征向量（这就丢失了空间信息）。以AlexNet为代表的经典CNN结构适合于图像级的分类和回归任务，因为它们最后都期望得到整个输入图像的一个数值描述（概率），比如AlexNet的ImageNet模型输出一个1000维的向量表示输入图像属于每一类的概率(softmax归一化)。\n",
    "\n",
    "但是 FCN 不同, FCN对图像进行像素级的分类，从而解决了语义级别的图像分割（semantic segmentation）问题。与经典的CNN在卷积层之后使用全连接层得到固定长度的特征向量进行分类（全联接层＋softmax输出）不同，**FCN可以接受任意尺寸的输入图像，采用反卷积层对最后一个卷积层的feature map进行上采样, 使它恢复到输入图像相同的尺寸**，从而可以对每个像素都产生了一个预测, 同时保留了原始输入图像中的空间信息, 最后在上采样的特征图上进行逐像素分类。\n",
    "\n",
    "\n",
    "总结一下就是:\n",
    "\n",
    "**CNN的输入是图像，输出是一个结果，或者说是一个值，一个概率值。**\n",
    "\n",
    "**FCN提出所追求的是，输入是一张图片是，输出也是一张图片，学习像素到像素的映射 。**\n",
    "\n",
    "这样的特征刚好符合语义分割任务中的,对像素进行分类.\n",
    "\n",
    "## 3. FCN的网络架构\n",
    "\n",
    "\n",
    "![network](./images/FCN_network.png)\n",
    "\n",
    "\n",
    "整个模型是 end-to-end 的.\n",
    "\n",
    "具体的流程是: 先**降采样（卷积、池化）**，再**上采样（反卷积）**\n",
    "\n",
    "降采样是为了更好的获得语义信息(因为不多用几层卷积就没法正确分类)。\n",
    "\n",
    "但是卷积和池化之后feature map变的很小了，需要还原成原图大小，所以就用上采样（反卷积）的方式把feature map 还原回原图大小。\n",
    "\n",
    "\n",
    "## 4. FCN中的关键技术\n",
    "\n",
    "### 4.1 反卷积(Deconvolution) 或者 转置卷积(conv_transpose)\n",
    "\n",
    "两者其实是等价的,只是称呼上有差异.\n",
    "\n",
    "详细的介绍就不做了,只给出一张比较直观地动态图\n",
    "\n",
    "2x2 的原始图像被还原成 4x4 的大小\n",
    "\n",
    "![conv_transpose](./images/decon.gif)\n",
    "\n",
    "\n",
    "### 4.2 FCN的前向编码过程\n",
    "\n",
    "就跟传统的CNN一样, Image(长为H,宽为W) 经过多个 conv 和一个 max-pooling 变成 pool1 feature,长宽变为 1/2 H 和 1/2 W;\n",
    "\n",
    "pool1 feature 再经过多个 conv 和一个 max-pooling 变成 pool2 feature,长宽变为 1/4 H 和 1/4 W;\n",
    "\n",
    "pool2 feature 再经过多个 conv 和一个 max-pooling 变成 pool3 feature,长宽变为 1/8 H 和 1/8 W;\n",
    "\n",
    "...\n",
    "\n",
    "\n",
    "一直到 pool5 feature, 它的长宽为 1/32 H 和 1/32 W;\n",
    "\n",
    "\n",
    "### 4.3 FCN的反向编码过程\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "![conv_transpose](./images/FCNdecon.png)\n",
    "\n",
    "对于 FCN-32s 而言,直接对 pool5 feature 进行32倍上采样获得 32 x upsampled feature, 再对 32 x upsampled feature 每个点做 softmax prediction 获得 32 x upsampled feature prediction (也就是分割图)\n",
    "\n",
    "对于 FCN-16s 而言,首先对 pool5 feature 进行2倍上采样获得 2 x upsampled feature, 再把 pool4 feature 和 2 x upsampled feature 逐点相加,然后对相加的 feature 进行16倍上采样,并做 softmax prediction 获得 16 x upsampled feature prediction\n",
    "\n",
    "对于 FCN-8s 而言,首先进行 pool4 feature 和 2 x upsampled feature 逐点相加,然后又进行 pool3 feature 和 2 x upsampled feature 逐点相加,即进行更多次的特征融合,具体形式如 16s.\n",
    "\n",
    "\n",
    "\n",
    "### 4.4 FCN 的总结\n",
    "\n",
    "1. 将寻常CNN最后几层的全连接层替换成卷几层,完成了针对像素的分类\n",
    "\n",
    "2. 阐述了CNN如何进行 end-to-end 的训练\n",
    "\n",
    "3. 输入图片的尺寸不再做限制,任意大小的输入图片都可以完成像素级别的分类(语义分割任务).\n",
    "\n",
    "## 5. 数据下载\n",
    "\n",
    "\n",
    "[PASCAL VOC2012数据集](http://host.robots.ox.ac.uk/pascal/VOC/voc2012/VOCtrainval_11-May-2012.tar)\n",
    "\n",
    "\n",
    "下载之后,解压文件夹,然后修改 voc_root 的值,比如,我解压到了上级目录的 data 文件夹中,那么我的 \n",
    "\n",
    "```\n",
    "voc_root = '../data/VOCdevkit/VOC2012'\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 导入需要的包\n",
    "import os\n",
    "import time\n",
    "import torch\n",
    "import warnings\n",
    "import numpy as np\n",
    "from torch import nn\n",
    "from PIL import Image\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "import torch.nn.functional as F\n",
    "from torch.autograd import Variable\n",
    "from torch.optim import lr_scheduler\n",
    "import torchvision.transforms as tfs\n",
    "from torchvision import datasets, models\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 数据路径,如果不一致,请自行修改\n",
    "voc_root = '../data/VOCdevkit/VOC2012'\n",
    "def read_images(root=voc_root, train=True):\n",
    "    txt_fname = root + '/ImageSets/Segmentation/' + ('train.txt' if train else 'val.txt')\n",
    "    with open(txt_fname, 'r') as f:\n",
    "        images = f.read().split()\n",
    "    data = [os.path.join(root, 'JPEGImages', i+'.jpg') for i in images]\n",
    "    label = [os.path.join(root, 'SegmentationClass', i+'.png') for i in images]\n",
    "    return data, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#等比例压缩图片\n",
    "def scale_high(img, target_high):\n",
    "    ow, oh = img.size\n",
    "    if (ow == target_high):\n",
    "        return img\n",
    "    h = target_high\n",
    "    w = int(target_high * ow / oh)\n",
    "    return img.resize((w, h), Image.BICUBIC)\n",
    "\n",
    "#中心裁剪\n",
    "def random_crop(data, label, crop_size):\n",
    "    transforms = tfs.CenterCrop(crop_size)\n",
    "    data = transforms(data)\n",
    "    label = transforms(label)\n",
    "    return data, label"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(21, 21)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "classes = ['background','aeroplane','bicycle','bird','boat',\n",
    "           'bottle','bus','car','cat','chair','cow','diningtable',\n",
    "           'dog','horse','motorbike','person','potted plant',\n",
    "           'sheep','sofa','train','tv/monitor']\n",
    "\n",
    "colormap = [[0,0,0],[128,0,0],[0,128,0], [128,128,0], [0,0,128],\n",
    "            [128,0,128],[0,128,128],[128,128,128],[64,0,0],[192,0,0],\n",
    "            [64,128,0],[192,128,0],[64,0,128],[192,0,128],\n",
    "            [64,128,128],[192,128,128],[0,64,0],[128,64,0],\n",
    "            [0,192,0],[128,192,0],[0,64,128]]\n",
    "\n",
    "len(classes), len(colormap)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 每个像素点有 0 ~ 255 的选择，RGB 三个通道\n",
    "cm2lbl = np.zeros(256**3) \n",
    "for i,cm in enumerate(colormap):\n",
    "    # 建立索引\n",
    "    cm2lbl[(cm[0]*256+cm[1])*256+cm[2]] = i \n",
    "\n",
    "def image2label(img):\n",
    "    data = np.array(img, dtype='int32')\n",
    "    idx = (data[:, :, 0] * 256 + data[:, :, 1]) * 256 + data[:, :, 2]\n",
    "    # 根据索引得到 label 矩阵\n",
    "    return np.array(cm2lbl[idx], dtype='int64') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def img_transforms(img, label, crop_size):\n",
    "    img, label = random_crop(img, label, crop_size)\n",
    "    img = scale_high(img,int(img.size[1]/2))\n",
    "    label = scale_high(label,int(label.size[1]/2))\n",
    "    img_tfs = tfs.Compose([      \n",
    "        tfs.ToTensor(),\n",
    "        tfs.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])\n",
    "    ])\n",
    "\n",
    "    img = img_tfs(img)\n",
    "    label = image2label(label)\n",
    "    label = torch.from_numpy(label)\n",
    "    return img, label\n",
    "\n",
    "#制作数据集\n",
    "class VOCSegDataset(Dataset):\n",
    "    def __init__(self, train, crop_size, transforms):\n",
    "        self.crop_size = crop_size\n",
    "        self.transforms = transforms\n",
    "        self.data_list, self.label_list = read_images(train=train)\n",
    "        print('Read ' + str(len(self.data_list)) + ' images')\n",
    "        \n",
    "        \n",
    "    def __getitem__(self, idx):\n",
    "        img = self.data_list[idx]\n",
    "        label = self.label_list[idx]\n",
    "        img = Image.open(img)\n",
    "        label = Image.open(label).convert('RGB')\n",
    "        img, label = self.transforms(img, label, self.crop_size)\n",
    "        return img, label\n",
    "    \n",
    "    def __len__(self):\n",
    "        return len(self.data_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Read 1464 images\n",
      "Read 1449 images\n"
     ]
    }
   ],
   "source": [
    "# 实例化数据集\n",
    "batch_size = 8\n",
    "input_shape = (512, 512)\n",
    "train_data = VOCSegDataset(True, input_shape, img_transforms)\n",
    "testdata = VOCSegDataset(False, input_shape, img_transforms)\n",
    "trainloader = DataLoader(train_data, batch_size = batch_size, shuffle=True)\n",
    "testloader = DataLoader(testdata, batch_size = batch_size, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#初始化权值\n",
    "def get_upsampling_weight(in_channels, out_channels, kernel_size):\n",
    "    \"\"\"Make a 2D bilinear kernel suitable for upsampling\"\"\"\n",
    "    factor = (kernel_size + 1) // 2\n",
    "    if kernel_size % 2 == 1:\n",
    "        center = factor - 1\n",
    "    else:\n",
    "        center = factor - 0.5\n",
    "    og = np.ogrid[:kernel_size, :kernel_size]\n",
    "    filt = (1 - abs(og[0] - center) / factor) * \\\n",
    "           (1 - abs(og[1] - center) / factor)\n",
    "    weight = np.zeros((in_channels, out_channels, kernel_size, kernel_size),\n",
    "                      dtype=np.float64)\n",
    "    weight[range(in_channels), range(out_channels), :, :] = filt\n",
    "    return torch.from_numpy(weight).float()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#FCN网络结构的定义\n",
    "n_class = len(classes)\n",
    "class FCN(nn.Module):\n",
    "    def __init__(self, n_class=21):\n",
    "        super(FCN, self).__init__()\n",
    "        # conv1\n",
    "        self.conv1_1 = nn.Conv2d(3, 64, 3, padding=100)\n",
    "        self.relu1_1 = nn.ReLU(inplace=True)\n",
    "        self.conv1_2 = nn.Conv2d(64, 64, 3, padding=1)\n",
    "        self.relu1_2 = nn.ReLU(inplace=True)\n",
    "        self.pool1 = nn.MaxPool2d(2, stride=2, ceil_mode=True)  # 1/2\n",
    "\n",
    "        # conv2\n",
    "        self.conv2_1 = nn.Conv2d(64, 128, 3, padding=1)\n",
    "        self.relu2_1 = nn.ReLU(inplace=True)\n",
    "        self.conv2_2 = nn.Conv2d(128, 128, 3, padding=1)\n",
    "        self.relu2_2 = nn.ReLU(inplace=True)\n",
    "        self.pool2 = nn.MaxPool2d(2, stride=2, ceil_mode=True)  # 1/4\n",
    "\n",
    "        # conv3\n",
    "        self.conv3_1 = nn.Conv2d(128, 256, 3, padding=1)\n",
    "        self.relu3_1 = nn.ReLU(inplace=True)\n",
    "        self.conv3_2 = nn.Conv2d(256, 256, 3, padding=1)\n",
    "        self.relu3_2 = nn.ReLU(inplace=True)\n",
    "        self.conv3_3 = nn.Conv2d(256, 256, 3, padding=1)\n",
    "        self.relu3_3 = nn.ReLU(inplace=True)\n",
    "        self.pool3 = nn.MaxPool2d(2, stride=2, ceil_mode=True)  # 1/8\n",
    "\n",
    "        # conv4\n",
    "        self.conv4_1 = nn.Conv2d(256, 512, 3, padding=1)\n",
    "        self.relu4_1 = nn.ReLU(inplace=True)\n",
    "        self.conv4_2 = nn.Conv2d(512, 512, 3, padding=1)\n",
    "        self.relu4_2 = nn.ReLU(inplace=True)\n",
    "        self.conv4_3 = nn.Conv2d(512, 512, 3, padding=1)\n",
    "        self.relu4_3 = nn.ReLU(inplace=True)\n",
    "        self.pool4 = nn.MaxPool2d(2, stride=2, ceil_mode=True)  # 1/16\n",
    "\n",
    "        # conv5\n",
    "        self.conv5_1 = nn.Conv2d(512, 512, 3, padding=1)\n",
    "        self.relu5_1 = nn.ReLU(inplace=True)\n",
    "        self.conv5_2 = nn.Conv2d(512, 512, 3, padding=1)\n",
    "        self.relu5_2 = nn.ReLU(inplace=True)\n",
    "        self.conv5_3 = nn.Conv2d(512, 512, 3, padding=1)\n",
    "        self.relu5_3 = nn.ReLU(inplace=True)\n",
    "        self.pool5 = nn.MaxPool2d(2, stride=2, ceil_mode=True)  # 1/32\n",
    "\n",
    "        # fc6\n",
    "        self.fc6 = nn.Conv2d(512, 4096, 7)\n",
    "        self.relu6 = nn.ReLU(inplace=True)\n",
    "        self.drop6 = nn.Dropout2d()\n",
    "\n",
    "        # fc7\n",
    "        self.fc7 = nn.Conv2d(4096, 4096, 1)\n",
    "        self.relu7 = nn.ReLU(inplace=True)\n",
    "        self.drop7 = nn.Dropout2d()\n",
    "\n",
    "        self.score_fr = nn.Conv2d(4096, n_class, 1)\n",
    "        self.score_pool3 = nn.Conv2d(256, n_class, 1)\n",
    "        self.score_pool4 = nn.Conv2d(512, n_class, 1)\n",
    "\n",
    "        self.upscore2 = nn.ConvTranspose2d(\n",
    "            n_class, n_class, 4, stride=2, bias=False)\n",
    "        self.upscore8 = nn.ConvTranspose2d(\n",
    "            n_class, n_class, 16, stride=8, bias=False)\n",
    "        self.upscore_pool4 = nn.ConvTranspose2d(\n",
    "            n_class, n_class, 4, stride=2, bias=False)\n",
    "\n",
    "        self._initialize_weights()\n",
    "\n",
    "    def _initialize_weights(self):\n",
    "        for m in self.modules():\n",
    "            if isinstance(m, nn.Conv2d):\n",
    "                m.weight.data.zero_()\n",
    "                if m.bias is not None:\n",
    "                    m.bias.data.zero_()\n",
    "            if isinstance(m, nn.ConvTranspose2d):\n",
    "                assert m.kernel_size[0] == m.kernel_size[1]\n",
    "                initial_weight = get_upsampling_weight(\n",
    "                    m.in_channels, m.out_channels, m.kernel_size[0])\n",
    "                m.weight.data.copy_(initial_weight)\n",
    "\n",
    "    def forward(self, x):\n",
    "        h = x\n",
    "        h = self.relu1_1(self.conv1_1(h))\n",
    "        h = self.relu1_2(self.conv1_2(h))\n",
    "        h = self.pool1(h)\n",
    "\n",
    "        h = self.relu2_1(self.conv2_1(h))\n",
    "        h = self.relu2_2(self.conv2_2(h))\n",
    "        h = self.pool2(h)\n",
    "\n",
    "        h = self.relu3_1(self.conv3_1(h))\n",
    "        h = self.relu3_2(self.conv3_2(h))\n",
    "        h = self.relu3_3(self.conv3_3(h))\n",
    "        h = self.pool3(h)\n",
    "        pool3 = h  # 1/8\n",
    "\n",
    "        h = self.relu4_1(self.conv4_1(h))\n",
    "        h = self.relu4_2(self.conv4_2(h))\n",
    "        h = self.relu4_3(self.conv4_3(h))\n",
    "        h = self.pool4(h)\n",
    "        pool4 = h  # 1/16\n",
    "\n",
    "        h = self.relu5_1(self.conv5_1(h))\n",
    "        h = self.relu5_2(self.conv5_2(h))\n",
    "        h = self.relu5_3(self.conv5_3(h))\n",
    "        h = self.pool5(h)\n",
    "\n",
    "        h = self.relu6(self.fc6(h))\n",
    "        h = self.drop6(h)\n",
    "\n",
    "        h = self.relu7(self.fc7(h))\n",
    "        h = self.drop7(h)\n",
    "\n",
    "        h = self.score_fr(h)\n",
    "        h = self.upscore2(h)\n",
    "        upscore2 = h  # 1/16\n",
    "\n",
    "        h = self.score_pool4(pool4)\n",
    "        h = h[:, :, 5:5 + upscore2.size()[2], 5:5 + upscore2.size()[3]]\n",
    "        score_pool4c = h  # 1/16\n",
    "\n",
    "        h = upscore2 + score_pool4c  # 1/16\n",
    "        h = self.upscore_pool4(h)\n",
    "        upscore_pool4 = h  # 1/8\n",
    "\n",
    "        h = self.score_pool3(pool3)\n",
    "        h = h[:, :,\n",
    "              9:9 + upscore_pool4.size()[2],\n",
    "              9:9 + upscore_pool4.size()[3]]\n",
    "        score_pool3c = h  # 1/8\n",
    "\n",
    "        h = upscore_pool4 + score_pool3c  # 1/8\n",
    "\n",
    "        h = self.upscore8(h)\n",
    "        h = h[:, :, 31:31 + x.size()[2], 31:31 + x.size()[3]].contiguous()\n",
    "\n",
    "        return h"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#模型实例化 \n",
    "model = FCN(n_class)\n",
    "use_gpu = False\n",
    "use_gpu = torch.cuda.is_available()\n",
    "if use_gpu:\n",
    "    model = model.cuda()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "FCN(\n",
       "  (conv1_1): Conv2d(3, 64, kernel_size=(3, 3), stride=(1, 1), padding=(100, 100))\n",
       "  (relu1_1): ReLU(inplace)\n",
       "  (conv1_2): Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu1_2): ReLU(inplace)\n",
       "  (pool1): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=True)\n",
       "  (conv2_1): Conv2d(64, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu2_1): ReLU(inplace)\n",
       "  (conv2_2): Conv2d(128, 128, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu2_2): ReLU(inplace)\n",
       "  (pool2): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=True)\n",
       "  (conv3_1): Conv2d(128, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu3_1): ReLU(inplace)\n",
       "  (conv3_2): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu3_2): ReLU(inplace)\n",
       "  (conv3_3): Conv2d(256, 256, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu3_3): ReLU(inplace)\n",
       "  (pool3): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=True)\n",
       "  (conv4_1): Conv2d(256, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu4_1): ReLU(inplace)\n",
       "  (conv4_2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu4_2): ReLU(inplace)\n",
       "  (conv4_3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu4_3): ReLU(inplace)\n",
       "  (pool4): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=True)\n",
       "  (conv5_1): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu5_1): ReLU(inplace)\n",
       "  (conv5_2): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu5_2): ReLU(inplace)\n",
       "  (conv5_3): Conv2d(512, 512, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1))\n",
       "  (relu5_3): ReLU(inplace)\n",
       "  (pool5): MaxPool2d(kernel_size=2, stride=2, padding=0, dilation=1, ceil_mode=True)\n",
       "  (fc6): Conv2d(512, 4096, kernel_size=(7, 7), stride=(1, 1))\n",
       "  (relu6): ReLU(inplace)\n",
       "  (drop6): Dropout2d(p=0.5)\n",
       "  (fc7): Conv2d(4096, 4096, kernel_size=(1, 1), stride=(1, 1))\n",
       "  (relu7): ReLU(inplace)\n",
       "  (drop7): Dropout2d(p=0.5)\n",
       "  (score_fr): Conv2d(4096, 21, kernel_size=(1, 1), stride=(1, 1))\n",
       "  (score_pool3): Conv2d(256, 21, kernel_size=(1, 1), stride=(1, 1))\n",
       "  (score_pool4): Conv2d(512, 21, kernel_size=(1, 1), stride=(1, 1))\n",
       "  (upscore2): ConvTranspose2d(21, 21, kernel_size=(4, 4), stride=(2, 2), bias=False)\n",
       "  (upscore8): ConvTranspose2d(21, 21, kernel_size=(16, 16), stride=(8, 8), bias=False)\n",
       "  (upscore_pool4): ConvTranspose2d(21, 21, kernel_size=(4, 4), stride=(2, 2), bias=False)\n",
       ")"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "num_epoches = 100 # 训练次数\n",
    "learning_rate = 0.001\n",
    "weight_decay=1e-4\n",
    "criterion = nn.NLLLoss2d()\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr=learning_rate, weight_decay=weight_decay)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# IoU的值定义：(候选区域)Region Proposal与（标记区域）Ground Truth的窗口的交集比并集的比值，如果IoU低于0.5，\n",
    "# 那么相当于目标还是没有检测到\n",
    "def _fast_hist(label_true, label_pred, n_class):\n",
    "    mask = (label_true >= 0) & (label_true < n_class)\n",
    "    hist = np.bincount(\n",
    "        n_class * label_true[mask].astype(int) +\n",
    "        label_pred[mask], minlength=n_class ** 2).reshape(n_class, n_class)\n",
    "    return hist\n",
    "\n",
    "def label_accuracy_score(label_trues, label_preds, n_class):\n",
    "    \"\"\"Returns accuracy score evaluation result.\n",
    "      - overall accuracy\n",
    "      - mean accuracy\n",
    "      - mean IU\n",
    "      - fwavacc\n",
    "    \"\"\"\n",
    "    hist = np.zeros((n_class, n_class))\n",
    "    for lt, lp in zip(label_trues, label_preds):\n",
    "        hist += _fast_hist(lt.flatten(), lp.flatten(), n_class)\n",
    "    acc = np.diag(hist).sum() / hist.sum()\n",
    "    acc_cls = np.diag(hist) / hist.sum(axis=1)\n",
    "    acc_cls = np.nanmean(acc_cls)\n",
    "    \n",
    "    iu = np.diag(hist) / (hist.sum(axis=1) + hist.sum(axis=0) - np.diag(hist))\n",
    "    mean_iu = np.nanmean(iu)\n",
    "    freq = hist.sum(axis=1) / hist.sum()\n",
    "    fwavacc = (freq[freq > 0] * iu[freq > 0]).sum()\n",
    "    return acc, acc_cls, mean_iu, fwavacc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#学习率的优化\n",
    "def exp_lr_scheduler(optimizer, epoch, init_lr=0.001, lr_decay_epoch=30):\n",
    "    \"\"\"Decay learning rate by a factor of 0.1 every lr_decay_epoch epochs.\"\"\"\n",
    "    lr = init_lr * (0.5**(epoch // lr_decay_epoch))\n",
    "\n",
    "    if epoch % lr_decay_epoch == 0:\n",
    "        print('LR is set to {}'.format(lr))\n",
    "\n",
    "    for param_group in optimizer.param_groups:\n",
    "        param_group['lr'] = lr\n",
    "\n",
    "    return optimizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#模型的训练与测试\n",
    "def train_test(optimizer):\n",
    "    #训练\n",
    "    epochacc_y=[]\n",
    "    wrong_all=[]\n",
    "    testacc_y=[]\n",
    "    epoch_y = []\n",
    "    for epoch in range(num_epoches):\n",
    "        print ('- - - - - epoch',epoch+1,'- - - - - - - -')\n",
    "        train_acc_cls = 0\n",
    "        epoch_mean_IoU = 0\n",
    "        train_fwavacc = 0\n",
    "        \n",
    "        batch_idx_sum = 0 \n",
    "        minibatch_loss=0\n",
    "        epoch_loss_sum=0\n",
    "        epoch_acc=0\n",
    "        \n",
    "        optimizer = exp_lr_scheduler(optimizer, epoch)\n",
    "        \n",
    "        for batch_idx, data in enumerate(trainloader , 1):\n",
    "            \n",
    "            batch_idx_sum += data[0].size()[0]\n",
    "            \n",
    "            if use_gpu:\n",
    "                im = Variable(data[0]).cuda()\n",
    "                label = Variable(data[1]).cuda()\n",
    "            else:\n",
    "                im = Variable(data[0])\n",
    "                label = Variable(data[1])\n",
    "            \n",
    "            out = model(im)\n",
    "\n",
    "            out = F.log_softmax(out) \n",
    "        \n",
    "            minibatch_loss = criterion(out, label)\n",
    "            \n",
    "            optimizer.zero_grad()\n",
    "            \n",
    "            minibatch_loss.backward()\n",
    "            \n",
    "            optimizer.step()\n",
    "            \n",
    "            # 该列表存放每一个batch的loss值\n",
    "            \n",
    "            epoch_loss_sum += minibatch_loss.data[0]\n",
    "\n",
    "            label_pred = out.max(dim=1)[1].data.cpu().numpy()\n",
    "            \n",
    "            label_true = label.data.cpu().numpy()\n",
    "            \n",
    "            for lbt, lbp in zip(label_true, label_pred):\n",
    "                acc, acc_cls, mean_iu, fwavacc = label_accuracy_score(lbt, lbp, n_class)\n",
    "                epoch_acc += acc\n",
    "                train_acc_cls += acc_cls\n",
    "                epoch_mean_IoU += mean_iu\n",
    "                train_fwavacc += fwavacc\n",
    "                        \n",
    "            if batch_idx % 20 == 0 :\n",
    "                print(\"Train Epoch: {} [{}/{} ({:.0f}%)] ,Loss: {:.6f} ,Accuracy:{:.2f}%, Train Mean IoU:{:.5f}\".format(\n",
    "                    epoch+1,\n",
    "                    batch_idx_sum,\n",
    "                    len(train_data),\n",
    "                    100. * batch_idx_sum / len(train_data),\n",
    "                    minibatch_loss.data[0],\n",
    "                    100.*epoch_acc / batch_idx_sum,    \n",
    "                    epoch_mean_IoU / batch_idx_sum     \n",
    "                    ))\n",
    "                \n",
    "        epoch_loss_average = epoch_loss_sum/len(trainloader)\n",
    "\n",
    "        epoch_y.append(epoch_loss_average)\n",
    "        \n",
    "        epoch_acc_average=epoch_acc/len(train_data)\n",
    "\n",
    "        epochacc_y.append(epoch_acc_average)\n",
    "    #############################################################\n",
    "    #测试\n",
    "        print ('- - - - - test stage- - - - - - - -')\n",
    "        model.eval()\n",
    "        step_sum = 0\n",
    "        test_loss = 0\n",
    "        test_acc = 0\n",
    "        test_acc_cls = 0\n",
    "        test_mean_IoU = 0\n",
    "        test_fwavacc = 0\n",
    "        for step, data in enumerate(testloader , 1):\n",
    "            \n",
    "            step_sum += data[0].size()[0]\n",
    "            \n",
    "            if use_gpu:\n",
    "                im = Variable(data[0], volatile=True).cuda()\n",
    "                label = Variable(data[1], volatile=True).cuda()\n",
    "            else:\n",
    "                im = Variable(data[0])\n",
    "                label = Variable(data[1])\n",
    "            \n",
    "            out = model(im)\n",
    "            out = F.log_softmax(out)\n",
    "            test_loss = criterion(out, label)\n",
    "\n",
    "            label_pred = out.max(dim=1)[1].data.cpu().numpy()\n",
    "            label_true = label.data.cpu().numpy()\n",
    "            \n",
    "            for lbt, lbp in zip(label_true, label_pred):\n",
    "                acc, acc_cls, mean_IoU, fwavacc = label_accuracy_score(lbt, lbp, n_class)\n",
    "                test_acc += acc\n",
    "                test_acc_cls += acc_cls\n",
    "                test_mean_IoU += mean_IoU\n",
    "                test_fwavacc += fwavacc\n",
    "        \n",
    "        testacc=test_acc/len(testloader)\n",
    "        \n",
    "        test_mean_IoU = test_mean_IoU/len(testloader)\n",
    "        \n",
    "        testacc_y.append(testacc)\n",
    "        \n",
    "        print(\"Test_accuracy: {:.2f}% , Test Mean IoU: {:.5f} \".format(testacc, test_mean_IoU))\n",
    "    return epoch_y, epochacc_y ,testacc_y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "- - - - - epoch 1 - - - - - - - -\n",
      "LR is set to 0.001\n",
      "Train Epoch: 1 [160/1464 (11%)] ,Loss: 2.402696 ,Accuracy:83.67%, Train Mean IoU:0.36136\n",
      "Train Epoch: 1 [320/1464 (22%)] ,Loss: 0.905181 ,Accuracy:84.18%, Train Mean IoU:0.36267\n",
      "Train Epoch: 1 [480/1464 (33%)] ,Loss: 0.856753 ,Accuracy:84.70%, Train Mean IoU:0.36606\n",
      "Train Epoch: 1 [640/1464 (44%)] ,Loss: 1.302709 ,Accuracy:84.56%, Train Mean IoU:0.36617\n",
      "Train Epoch: 1 [800/1464 (55%)] ,Loss: 0.969163 ,Accuracy:84.77%, Train Mean IoU:0.36792\n",
      "Train Epoch: 1 [960/1464 (66%)] ,Loss: 0.678338 ,Accuracy:84.86%, Train Mean IoU:0.36897\n",
      "Train Epoch: 1 [1120/1464 (77%)] ,Loss: 0.745183 ,Accuracy:84.85%, Train Mean IoU:0.36857\n",
      "Train Epoch: 1 [1280/1464 (87%)] ,Loss: 0.958804 ,Accuracy:84.87%, Train Mean IoU:0.36882\n",
      "Train Epoch: 1 [1440/1464 (98%)] ,Loss: 0.630206 ,Accuracy:84.89%, Train Mean IoU:0.36900\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 2 - - - - - - - -\n",
      "Train Epoch: 2 [160/1464 (11%)] ,Loss: 1.102825 ,Accuracy:83.70%, Train Mean IoU:0.37246\n",
      "Train Epoch: 2 [320/1464 (22%)] ,Loss: 0.810908 ,Accuracy:84.63%, Train Mean IoU:0.37265\n",
      "Train Epoch: 2 [480/1464 (33%)] ,Loss: 0.682829 ,Accuracy:84.53%, Train Mean IoU:0.36833\n",
      "Train Epoch: 2 [640/1464 (44%)] ,Loss: 0.695184 ,Accuracy:85.28%, Train Mean IoU:0.36868\n",
      "Train Epoch: 2 [800/1464 (55%)] ,Loss: 1.139562 ,Accuracy:85.09%, Train Mean IoU:0.36819\n",
      "Train Epoch: 2 [960/1464 (66%)] ,Loss: 1.001483 ,Accuracy:85.08%, Train Mean IoU:0.36840\n",
      "Train Epoch: 2 [1120/1464 (77%)] ,Loss: 1.091056 ,Accuracy:84.94%, Train Mean IoU:0.36807\n",
      "Train Epoch: 2 [1280/1464 (87%)] ,Loss: 0.499875 ,Accuracy:84.90%, Train Mean IoU:0.36889\n",
      "Train Epoch: 2 [1440/1464 (98%)] ,Loss: 0.739695 ,Accuracy:84.80%, Train Mean IoU:0.36827\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 3 - - - - - - - -\n",
      "Train Epoch: 3 [160/1464 (11%)] ,Loss: 0.710477 ,Accuracy:85.12%, Train Mean IoU:0.36769\n",
      "Train Epoch: 3 [320/1464 (22%)] ,Loss: 0.727313 ,Accuracy:85.75%, Train Mean IoU:0.37207\n",
      "Train Epoch: 3 [480/1464 (33%)] ,Loss: 0.857694 ,Accuracy:84.05%, Train Mean IoU:0.36376\n",
      "Train Epoch: 3 [640/1464 (44%)] ,Loss: 0.888654 ,Accuracy:84.62%, Train Mean IoU:0.36563\n",
      "Train Epoch: 3 [800/1464 (55%)] ,Loss: 1.087080 ,Accuracy:84.81%, Train Mean IoU:0.36798\n",
      "Train Epoch: 3 [960/1464 (66%)] ,Loss: 0.800249 ,Accuracy:85.03%, Train Mean IoU:0.36931\n",
      "Train Epoch: 3 [1120/1464 (77%)] ,Loss: 0.955998 ,Accuracy:84.88%, Train Mean IoU:0.36857\n",
      "Train Epoch: 3 [1280/1464 (87%)] ,Loss: 1.036914 ,Accuracy:85.01%, Train Mean IoU:0.36845\n",
      "Train Epoch: 3 [1440/1464 (98%)] ,Loss: 1.333202 ,Accuracy:84.92%, Train Mean IoU:0.36869\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 4 - - - - - - - -\n",
      "Train Epoch: 4 [160/1464 (11%)] ,Loss: 0.882636 ,Accuracy:83.99%, Train Mean IoU:0.37104\n",
      "Train Epoch: 4 [320/1464 (22%)] ,Loss: 0.781598 ,Accuracy:84.37%, Train Mean IoU:0.37392\n",
      "Train Epoch: 4 [480/1464 (33%)] ,Loss: 0.943671 ,Accuracy:83.76%, Train Mean IoU:0.36661\n",
      "Train Epoch: 4 [640/1464 (44%)] ,Loss: 0.777076 ,Accuracy:84.19%, Train Mean IoU:0.36492\n",
      "Train Epoch: 4 [800/1464 (55%)] ,Loss: 1.018301 ,Accuracy:84.39%, Train Mean IoU:0.36648\n",
      "Train Epoch: 4 [960/1464 (66%)] ,Loss: 0.884694 ,Accuracy:84.64%, Train Mean IoU:0.36720\n",
      "Train Epoch: 4 [1120/1464 (77%)] ,Loss: 0.569199 ,Accuracy:84.85%, Train Mean IoU:0.36932\n",
      "Train Epoch: 4 [1280/1464 (87%)] ,Loss: 0.764760 ,Accuracy:84.84%, Train Mean IoU:0.36792\n",
      "Train Epoch: 4 [1440/1464 (98%)] ,Loss: 0.676223 ,Accuracy:84.90%, Train Mean IoU:0.36868\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 5 - - - - - - - -\n",
      "Train Epoch: 5 [160/1464 (11%)] ,Loss: 1.210638 ,Accuracy:85.11%, Train Mean IoU:0.36572\n",
      "Train Epoch: 5 [320/1464 (22%)] ,Loss: 0.813625 ,Accuracy:84.16%, Train Mean IoU:0.36130\n",
      "Train Epoch: 5 [480/1464 (33%)] ,Loss: 0.976056 ,Accuracy:84.72%, Train Mean IoU:0.36308\n",
      "Train Epoch: 5 [640/1464 (44%)] ,Loss: 0.937845 ,Accuracy:84.66%, Train Mean IoU:0.36537\n",
      "Train Epoch: 5 [800/1464 (55%)] ,Loss: 1.027657 ,Accuracy:84.78%, Train Mean IoU:0.36533\n",
      "Train Epoch: 5 [960/1464 (66%)] ,Loss: 0.698131 ,Accuracy:85.01%, Train Mean IoU:0.36733\n",
      "Train Epoch: 5 [1120/1464 (77%)] ,Loss: 0.865015 ,Accuracy:84.89%, Train Mean IoU:0.36776\n",
      "Train Epoch: 5 [1280/1464 (87%)] ,Loss: 0.664083 ,Accuracy:85.04%, Train Mean IoU:0.36939\n",
      "Train Epoch: 5 [1440/1464 (98%)] ,Loss: 0.593903 ,Accuracy:85.01%, Train Mean IoU:0.36920\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 6 - - - - - - - -\n",
      "Train Epoch: 6 [160/1464 (11%)] ,Loss: 0.950869 ,Accuracy:85.20%, Train Mean IoU:0.37042\n",
      "Train Epoch: 6 [320/1464 (22%)] ,Loss: 0.895217 ,Accuracy:85.38%, Train Mean IoU:0.37325\n",
      "Train Epoch: 6 [480/1464 (33%)] ,Loss: 0.979094 ,Accuracy:85.03%, Train Mean IoU:0.37296\n",
      "Train Epoch: 6 [640/1464 (44%)] ,Loss: 0.990137 ,Accuracy:85.06%, Train Mean IoU:0.37261\n",
      "Train Epoch: 6 [800/1464 (55%)] ,Loss: 0.833549 ,Accuracy:85.00%, Train Mean IoU:0.37207\n",
      "Train Epoch: 6 [960/1464 (66%)] ,Loss: 0.843710 ,Accuracy:84.94%, Train Mean IoU:0.37017\n",
      "Train Epoch: 6 [1120/1464 (77%)] ,Loss: 0.449317 ,Accuracy:85.01%, Train Mean IoU:0.37034\n",
      "Train Epoch: 6 [1280/1464 (87%)] ,Loss: 0.796198 ,Accuracy:85.00%, Train Mean IoU:0.36968\n",
      "Train Epoch: 6 [1440/1464 (98%)] ,Loss: 0.930101 ,Accuracy:84.88%, Train Mean IoU:0.36864\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 7 - - - - - - - -\n",
      "Train Epoch: 7 [160/1464 (11%)] ,Loss: 0.540507 ,Accuracy:85.12%, Train Mean IoU:0.37277\n",
      "Train Epoch: 7 [320/1464 (22%)] ,Loss: 0.870629 ,Accuracy:83.76%, Train Mean IoU:0.36375\n",
      "Train Epoch: 7 [480/1464 (33%)] ,Loss: 0.513141 ,Accuracy:84.64%, Train Mean IoU:0.36912\n",
      "Train Epoch: 7 [640/1464 (44%)] ,Loss: 0.890994 ,Accuracy:84.96%, Train Mean IoU:0.36862\n",
      "Train Epoch: 7 [800/1464 (55%)] ,Loss: 0.598856 ,Accuracy:85.04%, Train Mean IoU:0.36602\n",
      "Train Epoch: 7 [960/1464 (66%)] ,Loss: 0.736288 ,Accuracy:84.82%, Train Mean IoU:0.36681\n",
      "Train Epoch: 7 [1120/1464 (77%)] ,Loss: 0.712744 ,Accuracy:84.93%, Train Mean IoU:0.36744\n",
      "Train Epoch: 7 [1280/1464 (87%)] ,Loss: 1.102563 ,Accuracy:85.03%, Train Mean IoU:0.36844\n",
      "Train Epoch: 7 [1440/1464 (98%)] ,Loss: 0.782224 ,Accuracy:84.97%, Train Mean IoU:0.36924\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 8 - - - - - - - -\n",
      "Train Epoch: 8 [160/1464 (11%)] ,Loss: 0.737369 ,Accuracy:83.58%, Train Mean IoU:0.35823\n",
      "Train Epoch: 8 [320/1464 (22%)] ,Loss: 0.862385 ,Accuracy:84.23%, Train Mean IoU:0.36520\n",
      "Train Epoch: 8 [480/1464 (33%)] ,Loss: 0.690944 ,Accuracy:84.67%, Train Mean IoU:0.36805\n",
      "Train Epoch: 8 [640/1464 (44%)] ,Loss: 0.739472 ,Accuracy:84.82%, Train Mean IoU:0.37013\n",
      "Train Epoch: 8 [800/1464 (55%)] ,Loss: 0.664188 ,Accuracy:84.68%, Train Mean IoU:0.36709\n",
      "Train Epoch: 8 [960/1464 (66%)] ,Loss: 0.610187 ,Accuracy:84.77%, Train Mean IoU:0.36669\n",
      "Train Epoch: 8 [1120/1464 (77%)] ,Loss: 0.946037 ,Accuracy:84.75%, Train Mean IoU:0.36690\n",
      "Train Epoch: 8 [1280/1464 (87%)] ,Loss: 0.760717 ,Accuracy:84.79%, Train Mean IoU:0.36773\n",
      "Train Epoch: 8 [1440/1464 (98%)] ,Loss: 0.943452 ,Accuracy:84.86%, Train Mean IoU:0.36829\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 9 - - - - - - - -\n",
      "Train Epoch: 9 [160/1464 (11%)] ,Loss: 0.923526 ,Accuracy:86.94%, Train Mean IoU:0.37569\n",
      "Train Epoch: 9 [320/1464 (22%)] ,Loss: 1.103103 ,Accuracy:84.72%, Train Mean IoU:0.36690\n",
      "Train Epoch: 9 [480/1464 (33%)] ,Loss: 0.761520 ,Accuracy:85.45%, Train Mean IoU:0.37349\n",
      "Train Epoch: 9 [640/1464 (44%)] ,Loss: 0.833617 ,Accuracy:85.71%, Train Mean IoU:0.37485\n",
      "Train Epoch: 9 [800/1464 (55%)] ,Loss: 0.545286 ,Accuracy:85.77%, Train Mean IoU:0.37574\n",
      "Train Epoch: 9 [960/1464 (66%)] ,Loss: 0.801428 ,Accuracy:85.61%, Train Mean IoU:0.37483\n",
      "Train Epoch: 9 [1120/1464 (77%)] ,Loss: 0.660519 ,Accuracy:85.47%, Train Mean IoU:0.37265\n",
      "Train Epoch: 9 [1280/1464 (87%)] ,Loss: 0.991546 ,Accuracy:85.20%, Train Mean IoU:0.37048\n",
      "Train Epoch: 9 [1440/1464 (98%)] ,Loss: 0.955177 ,Accuracy:84.95%, Train Mean IoU:0.36907\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 10 - - - - - - - -\n",
      "Train Epoch: 10 [160/1464 (11%)] ,Loss: 0.834782 ,Accuracy:85.24%, Train Mean IoU:0.36799\n",
      "Train Epoch: 10 [320/1464 (22%)] ,Loss: 0.776323 ,Accuracy:84.58%, Train Mean IoU:0.36405\n",
      "Train Epoch: 10 [480/1464 (33%)] ,Loss: 1.175837 ,Accuracy:84.94%, Train Mean IoU:0.36709\n",
      "Train Epoch: 10 [640/1464 (44%)] ,Loss: 0.743804 ,Accuracy:84.42%, Train Mean IoU:0.36539\n",
      "Train Epoch: 10 [800/1464 (55%)] ,Loss: 0.849884 ,Accuracy:84.78%, Train Mean IoU:0.36730\n",
      "Train Epoch: 10 [960/1464 (66%)] ,Loss: 0.607467 ,Accuracy:84.85%, Train Mean IoU:0.36829\n",
      "Train Epoch: 10 [1120/1464 (77%)] ,Loss: 0.420894 ,Accuracy:84.89%, Train Mean IoU:0.36887\n",
      "Train Epoch: 10 [1280/1464 (87%)] ,Loss: 0.675807 ,Accuracy:84.98%, Train Mean IoU:0.36923\n",
      "Train Epoch: 10 [1440/1464 (98%)] ,Loss: 0.834539 ,Accuracy:84.86%, Train Mean IoU:0.36828\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 11 - - - - - - - -\n",
      "Train Epoch: 11 [160/1464 (11%)] ,Loss: 0.714221 ,Accuracy:87.12%, Train Mean IoU:0.37514\n",
      "Train Epoch: 11 [320/1464 (22%)] ,Loss: 0.840642 ,Accuracy:86.27%, Train Mean IoU:0.37012\n",
      "Train Epoch: 11 [480/1464 (33%)] ,Loss: 0.792832 ,Accuracy:85.90%, Train Mean IoU:0.37298\n",
      "Train Epoch: 11 [640/1464 (44%)] ,Loss: 0.927955 ,Accuracy:85.35%, Train Mean IoU:0.37161\n",
      "Train Epoch: 11 [800/1464 (55%)] ,Loss: 0.751268 ,Accuracy:85.05%, Train Mean IoU:0.36967\n",
      "Train Epoch: 11 [960/1464 (66%)] ,Loss: 0.737588 ,Accuracy:85.31%, Train Mean IoU:0.37133\n",
      "Train Epoch: 11 [1120/1464 (77%)] ,Loss: 0.760455 ,Accuracy:84.98%, Train Mean IoU:0.36932\n",
      "Train Epoch: 11 [1280/1464 (87%)] ,Loss: 1.416958 ,Accuracy:84.83%, Train Mean IoU:0.36718\n",
      "Train Epoch: 11 [1440/1464 (98%)] ,Loss: 0.744000 ,Accuracy:84.88%, Train Mean IoU:0.36828\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 12 - - - - - - - -\n",
      "Train Epoch: 12 [160/1464 (11%)] ,Loss: 1.050759 ,Accuracy:84.14%, Train Mean IoU:0.35616\n",
      "Train Epoch: 12 [320/1464 (22%)] ,Loss: 0.744612 ,Accuracy:84.77%, Train Mean IoU:0.37147\n",
      "Train Epoch: 12 [480/1464 (33%)] ,Loss: 0.873615 ,Accuracy:84.84%, Train Mean IoU:0.37088\n",
      "Train Epoch: 12 [640/1464 (44%)] ,Loss: 0.760835 ,Accuracy:84.69%, Train Mean IoU:0.37124\n",
      "Train Epoch: 12 [800/1464 (55%)] ,Loss: 0.698656 ,Accuracy:84.75%, Train Mean IoU:0.36824\n",
      "Train Epoch: 12 [960/1464 (66%)] ,Loss: 0.960511 ,Accuracy:84.78%, Train Mean IoU:0.36817\n",
      "Train Epoch: 12 [1120/1464 (77%)] ,Loss: 1.233918 ,Accuracy:84.90%, Train Mean IoU:0.36791\n",
      "Train Epoch: 12 [1280/1464 (87%)] ,Loss: 0.810618 ,Accuracy:84.97%, Train Mean IoU:0.36974\n",
      "Train Epoch: 12 [1440/1464 (98%)] ,Loss: 1.343855 ,Accuracy:84.88%, Train Mean IoU:0.36923\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 13 - - - - - - - -\n",
      "Train Epoch: 13 [160/1464 (11%)] ,Loss: 0.738661 ,Accuracy:84.04%, Train Mean IoU:0.35440\n",
      "Train Epoch: 13 [320/1464 (22%)] ,Loss: 0.696547 ,Accuracy:84.23%, Train Mean IoU:0.36258\n",
      "Train Epoch: 13 [480/1464 (33%)] ,Loss: 0.681523 ,Accuracy:84.46%, Train Mean IoU:0.36419\n",
      "Train Epoch: 13 [640/1464 (44%)] ,Loss: 0.915831 ,Accuracy:84.90%, Train Mean IoU:0.36806\n",
      "Train Epoch: 13 [800/1464 (55%)] ,Loss: 0.981726 ,Accuracy:84.82%, Train Mean IoU:0.36854\n",
      "Train Epoch: 13 [960/1464 (66%)] ,Loss: 0.860459 ,Accuracy:84.90%, Train Mean IoU:0.36994\n",
      "Train Epoch: 13 [1120/1464 (77%)] ,Loss: 1.063385 ,Accuracy:84.92%, Train Mean IoU:0.36964\n",
      "Train Epoch: 13 [1280/1464 (87%)] ,Loss: 1.132265 ,Accuracy:84.90%, Train Mean IoU:0.36862\n",
      "Train Epoch: 13 [1440/1464 (98%)] ,Loss: 1.026113 ,Accuracy:84.94%, Train Mean IoU:0.36843\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 14 - - - - - - - -\n",
      "Train Epoch: 14 [160/1464 (11%)] ,Loss: 0.927118 ,Accuracy:83.61%, Train Mean IoU:0.36323\n",
      "Train Epoch: 14 [320/1464 (22%)] ,Loss: 0.541899 ,Accuracy:84.92%, Train Mean IoU:0.36685\n",
      "Train Epoch: 14 [480/1464 (33%)] ,Loss: 0.869474 ,Accuracy:84.85%, Train Mean IoU:0.36942\n",
      "Train Epoch: 14 [640/1464 (44%)] ,Loss: 0.810748 ,Accuracy:84.57%, Train Mean IoU:0.36708\n",
      "Train Epoch: 14 [800/1464 (55%)] ,Loss: 0.788884 ,Accuracy:84.75%, Train Mean IoU:0.36667\n",
      "Train Epoch: 14 [960/1464 (66%)] ,Loss: 0.703077 ,Accuracy:84.89%, Train Mean IoU:0.36796\n",
      "Train Epoch: 14 [1120/1464 (77%)] ,Loss: 1.106790 ,Accuracy:84.98%, Train Mean IoU:0.36807\n",
      "Train Epoch: 14 [1280/1464 (87%)] ,Loss: 1.315602 ,Accuracy:85.03%, Train Mean IoU:0.36817\n",
      "Train Epoch: 14 [1440/1464 (98%)] ,Loss: 1.224853 ,Accuracy:84.89%, Train Mean IoU:0.36870\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 15 - - - - - - - -\n",
      "Train Epoch: 15 [160/1464 (11%)] ,Loss: 1.015152 ,Accuracy:85.21%, Train Mean IoU:0.36664\n",
      "Train Epoch: 15 [320/1464 (22%)] ,Loss: 1.194335 ,Accuracy:84.94%, Train Mean IoU:0.36721\n",
      "Train Epoch: 15 [480/1464 (33%)] ,Loss: 0.624881 ,Accuracy:85.49%, Train Mean IoU:0.37025\n",
      "Train Epoch: 15 [640/1464 (44%)] ,Loss: 0.844624 ,Accuracy:84.82%, Train Mean IoU:0.36929\n",
      "Train Epoch: 15 [800/1464 (55%)] ,Loss: 0.720411 ,Accuracy:84.80%, Train Mean IoU:0.36709\n",
      "Train Epoch: 15 [960/1464 (66%)] ,Loss: 0.612630 ,Accuracy:84.93%, Train Mean IoU:0.36744\n",
      "Train Epoch: 15 [1120/1464 (77%)] ,Loss: 0.901497 ,Accuracy:84.86%, Train Mean IoU:0.36697\n",
      "Train Epoch: 15 [1280/1464 (87%)] ,Loss: 1.039563 ,Accuracy:84.81%, Train Mean IoU:0.36760\n",
      "Train Epoch: 15 [1440/1464 (98%)] ,Loss: 0.866698 ,Accuracy:84.85%, Train Mean IoU:0.36876\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 16 - - - - - - - -\n",
      "Train Epoch: 16 [160/1464 (11%)] ,Loss: 0.992523 ,Accuracy:84.06%, Train Mean IoU:0.35444\n",
      "Train Epoch: 16 [320/1464 (22%)] ,Loss: 0.939295 ,Accuracy:84.14%, Train Mean IoU:0.36602\n",
      "Train Epoch: 16 [480/1464 (33%)] ,Loss: 0.845067 ,Accuracy:84.83%, Train Mean IoU:0.36747\n",
      "Train Epoch: 16 [640/1464 (44%)] ,Loss: 0.958922 ,Accuracy:85.32%, Train Mean IoU:0.37263\n",
      "Train Epoch: 16 [800/1464 (55%)] ,Loss: 0.484671 ,Accuracy:85.22%, Train Mean IoU:0.37189\n",
      "Train Epoch: 16 [960/1464 (66%)] ,Loss: 0.681365 ,Accuracy:85.19%, Train Mean IoU:0.37196\n",
      "Train Epoch: 16 [1120/1464 (77%)] ,Loss: 0.926144 ,Accuracy:85.05%, Train Mean IoU:0.36977\n",
      "Train Epoch: 16 [1280/1464 (87%)] ,Loss: 0.839970 ,Accuracy:85.10%, Train Mean IoU:0.36942\n",
      "Train Epoch: 16 [1440/1464 (98%)] ,Loss: 1.231076 ,Accuracy:84.94%, Train Mean IoU:0.36899\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 17 - - - - - - - -\n",
      "Train Epoch: 17 [160/1464 (11%)] ,Loss: 1.217800 ,Accuracy:84.93%, Train Mean IoU:0.36749\n",
      "Train Epoch: 17 [320/1464 (22%)] ,Loss: 0.877135 ,Accuracy:84.99%, Train Mean IoU:0.37508\n",
      "Train Epoch: 17 [480/1464 (33%)] ,Loss: 1.125367 ,Accuracy:85.03%, Train Mean IoU:0.37112\n",
      "Train Epoch: 17 [640/1464 (44%)] ,Loss: 0.838024 ,Accuracy:84.99%, Train Mean IoU:0.37092\n",
      "Train Epoch: 17 [800/1464 (55%)] ,Loss: 0.882648 ,Accuracy:85.18%, Train Mean IoU:0.37212\n",
      "Train Epoch: 17 [960/1464 (66%)] ,Loss: 0.891513 ,Accuracy:85.07%, Train Mean IoU:0.37105\n",
      "Train Epoch: 17 [1120/1464 (77%)] ,Loss: 0.936474 ,Accuracy:84.96%, Train Mean IoU:0.37076\n",
      "Train Epoch: 17 [1280/1464 (87%)] ,Loss: 0.799077 ,Accuracy:85.01%, Train Mean IoU:0.36925\n",
      "Train Epoch: 17 [1440/1464 (98%)] ,Loss: 0.931270 ,Accuracy:84.90%, Train Mean IoU:0.36886\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 18 - - - - - - - -\n",
      "Train Epoch: 18 [160/1464 (11%)] ,Loss: 0.958331 ,Accuracy:84.45%, Train Mean IoU:0.36452\n",
      "Train Epoch: 18 [320/1464 (22%)] ,Loss: 1.039423 ,Accuracy:83.63%, Train Mean IoU:0.36170\n",
      "Train Epoch: 18 [480/1464 (33%)] ,Loss: 1.225535 ,Accuracy:84.22%, Train Mean IoU:0.36718\n",
      "Train Epoch: 18 [640/1464 (44%)] ,Loss: 0.733809 ,Accuracy:84.37%, Train Mean IoU:0.36828\n",
      "Train Epoch: 18 [800/1464 (55%)] ,Loss: 0.712676 ,Accuracy:84.58%, Train Mean IoU:0.36946\n",
      "Train Epoch: 18 [960/1464 (66%)] ,Loss: 0.697234 ,Accuracy:84.39%, Train Mean IoU:0.36772\n",
      "Train Epoch: 18 [1120/1464 (77%)] ,Loss: 1.285867 ,Accuracy:84.67%, Train Mean IoU:0.36857\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 18 [1280/1464 (87%)] ,Loss: 0.926769 ,Accuracy:84.94%, Train Mean IoU:0.36976\n",
      "Train Epoch: 18 [1440/1464 (98%)] ,Loss: 0.960551 ,Accuracy:84.87%, Train Mean IoU:0.36849\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 19 - - - - - - - -\n",
      "Train Epoch: 19 [160/1464 (11%)] ,Loss: 0.852322 ,Accuracy:86.32%, Train Mean IoU:0.38413\n",
      "Train Epoch: 19 [320/1464 (22%)] ,Loss: 0.658716 ,Accuracy:84.84%, Train Mean IoU:0.36890\n",
      "Train Epoch: 19 [480/1464 (33%)] ,Loss: 0.626228 ,Accuracy:85.28%, Train Mean IoU:0.36765\n",
      "Train Epoch: 19 [640/1464 (44%)] ,Loss: 0.751829 ,Accuracy:85.28%, Train Mean IoU:0.36994\n",
      "Train Epoch: 19 [800/1464 (55%)] ,Loss: 0.751632 ,Accuracy:85.07%, Train Mean IoU:0.36963\n",
      "Train Epoch: 19 [960/1464 (66%)] ,Loss: 0.639441 ,Accuracy:84.99%, Train Mean IoU:0.37047\n",
      "Train Epoch: 19 [1120/1464 (77%)] ,Loss: 0.732789 ,Accuracy:84.92%, Train Mean IoU:0.36911\n",
      "Train Epoch: 19 [1280/1464 (87%)] ,Loss: 0.920148 ,Accuracy:84.68%, Train Mean IoU:0.36766\n",
      "Train Epoch: 19 [1440/1464 (98%)] ,Loss: 0.808752 ,Accuracy:84.85%, Train Mean IoU:0.36862\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 20 - - - - - - - -\n",
      "Train Epoch: 20 [160/1464 (11%)] ,Loss: 0.837800 ,Accuracy:84.61%, Train Mean IoU:0.36615\n",
      "Train Epoch: 20 [320/1464 (22%)] ,Loss: 1.105462 ,Accuracy:84.46%, Train Mean IoU:0.36571\n",
      "Train Epoch: 20 [480/1464 (33%)] ,Loss: 0.958280 ,Accuracy:84.59%, Train Mean IoU:0.36781\n",
      "Train Epoch: 20 [640/1464 (44%)] ,Loss: 0.774321 ,Accuracy:84.58%, Train Mean IoU:0.36740\n",
      "Train Epoch: 20 [800/1464 (55%)] ,Loss: 1.021924 ,Accuracy:84.83%, Train Mean IoU:0.36748\n",
      "Train Epoch: 20 [960/1464 (66%)] ,Loss: 0.676041 ,Accuracy:84.94%, Train Mean IoU:0.36926\n",
      "Train Epoch: 20 [1120/1464 (77%)] ,Loss: 0.575012 ,Accuracy:84.86%, Train Mean IoU:0.36884\n",
      "Train Epoch: 20 [1280/1464 (87%)] ,Loss: 0.613178 ,Accuracy:84.76%, Train Mean IoU:0.36786\n",
      "Train Epoch: 20 [1440/1464 (98%)] ,Loss: 0.690859 ,Accuracy:84.92%, Train Mean IoU:0.36871\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 21 - - - - - - - -\n",
      "Train Epoch: 21 [160/1464 (11%)] ,Loss: 0.676804 ,Accuracy:85.33%, Train Mean IoU:0.37513\n",
      "Train Epoch: 21 [320/1464 (22%)] ,Loss: 1.414255 ,Accuracy:85.16%, Train Mean IoU:0.36998\n",
      "Train Epoch: 21 [480/1464 (33%)] ,Loss: 0.740850 ,Accuracy:85.08%, Train Mean IoU:0.36955\n",
      "Train Epoch: 21 [640/1464 (44%)] ,Loss: 0.576715 ,Accuracy:84.82%, Train Mean IoU:0.36985\n",
      "Train Epoch: 21 [800/1464 (55%)] ,Loss: 0.909887 ,Accuracy:84.48%, Train Mean IoU:0.36631\n",
      "Train Epoch: 21 [960/1464 (66%)] ,Loss: 0.883613 ,Accuracy:84.82%, Train Mean IoU:0.36866\n",
      "Train Epoch: 21 [1120/1464 (77%)] ,Loss: 0.616557 ,Accuracy:84.95%, Train Mean IoU:0.36939\n",
      "Train Epoch: 21 [1280/1464 (87%)] ,Loss: 1.193419 ,Accuracy:84.76%, Train Mean IoU:0.36811\n",
      "Train Epoch: 21 [1440/1464 (98%)] ,Loss: 0.619617 ,Accuracy:84.85%, Train Mean IoU:0.36871\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 22 - - - - - - - -\n",
      "Train Epoch: 22 [160/1464 (11%)] ,Loss: 0.845579 ,Accuracy:84.95%, Train Mean IoU:0.34954\n",
      "Train Epoch: 22 [320/1464 (22%)] ,Loss: 0.830331 ,Accuracy:85.50%, Train Mean IoU:0.36489\n",
      "Train Epoch: 22 [480/1464 (33%)] ,Loss: 1.081616 ,Accuracy:85.05%, Train Mean IoU:0.36443\n",
      "Train Epoch: 22 [640/1464 (44%)] ,Loss: 0.888319 ,Accuracy:85.06%, Train Mean IoU:0.36737\n",
      "Train Epoch: 22 [800/1464 (55%)] ,Loss: 0.746459 ,Accuracy:84.85%, Train Mean IoU:0.36644\n",
      "Train Epoch: 22 [960/1464 (66%)] ,Loss: 1.148504 ,Accuracy:84.83%, Train Mean IoU:0.36662\n",
      "Train Epoch: 22 [1120/1464 (77%)] ,Loss: 0.716593 ,Accuracy:84.74%, Train Mean IoU:0.36767\n",
      "Train Epoch: 22 [1280/1464 (87%)] ,Loss: 0.676054 ,Accuracy:84.79%, Train Mean IoU:0.36873\n",
      "Train Epoch: 22 [1440/1464 (98%)] ,Loss: 0.764152 ,Accuracy:85.01%, Train Mean IoU:0.36941\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 23 - - - - - - - -\n",
      "Train Epoch: 23 [160/1464 (11%)] ,Loss: 1.115494 ,Accuracy:86.91%, Train Mean IoU:0.38025\n",
      "Train Epoch: 23 [320/1464 (22%)] ,Loss: 0.984360 ,Accuracy:85.08%, Train Mean IoU:0.37244\n",
      "Train Epoch: 23 [480/1464 (33%)] ,Loss: 0.667921 ,Accuracy:84.87%, Train Mean IoU:0.37143\n",
      "Train Epoch: 23 [640/1464 (44%)] ,Loss: 0.784030 ,Accuracy:85.07%, Train Mean IoU:0.36929\n",
      "Train Epoch: 23 [800/1464 (55%)] ,Loss: 1.252862 ,Accuracy:84.87%, Train Mean IoU:0.36958\n",
      "Train Epoch: 23 [960/1464 (66%)] ,Loss: 0.989014 ,Accuracy:84.99%, Train Mean IoU:0.36855\n",
      "Train Epoch: 23 [1120/1464 (77%)] ,Loss: 0.935206 ,Accuracy:84.91%, Train Mean IoU:0.36908\n",
      "Train Epoch: 23 [1280/1464 (87%)] ,Loss: 0.857265 ,Accuracy:84.93%, Train Mean IoU:0.36981\n",
      "Train Epoch: 23 [1440/1464 (98%)] ,Loss: 0.717078 ,Accuracy:84.98%, Train Mean IoU:0.36925\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 24 - - - - - - - -\n",
      "Train Epoch: 24 [160/1464 (11%)] ,Loss: 0.898857 ,Accuracy:85.73%, Train Mean IoU:0.37567\n",
      "Train Epoch: 24 [320/1464 (22%)] ,Loss: 0.941885 ,Accuracy:85.45%, Train Mean IoU:0.37214\n",
      "Train Epoch: 24 [480/1464 (33%)] ,Loss: 0.656379 ,Accuracy:84.96%, Train Mean IoU:0.36898\n",
      "Train Epoch: 24 [640/1464 (44%)] ,Loss: 1.057352 ,Accuracy:85.09%, Train Mean IoU:0.37108\n",
      "Train Epoch: 24 [800/1464 (55%)] ,Loss: 0.739037 ,Accuracy:84.68%, Train Mean IoU:0.36874\n",
      "Train Epoch: 24 [960/1464 (66%)] ,Loss: 0.632728 ,Accuracy:84.73%, Train Mean IoU:0.36829\n",
      "Train Epoch: 24 [1120/1464 (77%)] ,Loss: 0.762531 ,Accuracy:84.72%, Train Mean IoU:0.36844\n",
      "Train Epoch: 24 [1280/1464 (87%)] ,Loss: 0.984917 ,Accuracy:84.70%, Train Mean IoU:0.36771\n",
      "Train Epoch: 24 [1440/1464 (98%)] ,Loss: 0.819411 ,Accuracy:84.81%, Train Mean IoU:0.36841\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 25 - - - - - - - -\n",
      "Train Epoch: 25 [160/1464 (11%)] ,Loss: 0.888035 ,Accuracy:82.59%, Train Mean IoU:0.36602\n",
      "Train Epoch: 25 [320/1464 (22%)] ,Loss: 0.876687 ,Accuracy:84.10%, Train Mean IoU:0.36820\n",
      "Train Epoch: 25 [480/1464 (33%)] ,Loss: 1.027701 ,Accuracy:84.98%, Train Mean IoU:0.36948\n",
      "Train Epoch: 25 [640/1464 (44%)] ,Loss: 0.617828 ,Accuracy:84.94%, Train Mean IoU:0.36902\n",
      "Train Epoch: 25 [800/1464 (55%)] ,Loss: 0.625843 ,Accuracy:85.17%, Train Mean IoU:0.37048\n",
      "Train Epoch: 25 [960/1464 (66%)] ,Loss: 0.821577 ,Accuracy:85.07%, Train Mean IoU:0.36871\n",
      "Train Epoch: 25 [1120/1464 (77%)] ,Loss: 1.257864 ,Accuracy:85.04%, Train Mean IoU:0.36917\n",
      "Train Epoch: 25 [1280/1464 (87%)] ,Loss: 1.077660 ,Accuracy:84.74%, Train Mean IoU:0.36769\n",
      "Train Epoch: 25 [1440/1464 (98%)] ,Loss: 0.529615 ,Accuracy:84.83%, Train Mean IoU:0.36832\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 26 - - - - - - - -\n",
      "Train Epoch: 26 [160/1464 (11%)] ,Loss: 0.848700 ,Accuracy:85.07%, Train Mean IoU:0.37175\n",
      "Train Epoch: 26 [320/1464 (22%)] ,Loss: 0.616136 ,Accuracy:85.21%, Train Mean IoU:0.36893\n",
      "Train Epoch: 26 [480/1464 (33%)] ,Loss: 0.810363 ,Accuracy:85.11%, Train Mean IoU:0.37165\n",
      "Train Epoch: 26 [640/1464 (44%)] ,Loss: 0.577244 ,Accuracy:85.33%, Train Mean IoU:0.37218\n",
      "Train Epoch: 26 [800/1464 (55%)] ,Loss: 1.017164 ,Accuracy:85.22%, Train Mean IoU:0.37080\n",
      "Train Epoch: 26 [960/1464 (66%)] ,Loss: 0.898563 ,Accuracy:85.06%, Train Mean IoU:0.37026\n",
      "Train Epoch: 26 [1120/1464 (77%)] ,Loss: 0.892481 ,Accuracy:85.17%, Train Mean IoU:0.37048\n",
      "Train Epoch: 26 [1280/1464 (87%)] ,Loss: 0.828281 ,Accuracy:85.04%, Train Mean IoU:0.36975\n",
      "Train Epoch: 26 [1440/1464 (98%)] ,Loss: 0.920282 ,Accuracy:84.96%, Train Mean IoU:0.36925\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 27 - - - - - - - -\n",
      "Train Epoch: 27 [160/1464 (11%)] ,Loss: 1.104995 ,Accuracy:85.40%, Train Mean IoU:0.37256\n",
      "Train Epoch: 27 [320/1464 (22%)] ,Loss: 0.799946 ,Accuracy:84.80%, Train Mean IoU:0.36878\n",
      "Train Epoch: 27 [480/1464 (33%)] ,Loss: 0.880104 ,Accuracy:84.85%, Train Mean IoU:0.36856\n",
      "Train Epoch: 27 [640/1464 (44%)] ,Loss: 0.662763 ,Accuracy:85.23%, Train Mean IoU:0.36998\n",
      "Train Epoch: 27 [800/1464 (55%)] ,Loss: 0.607669 ,Accuracy:85.17%, Train Mean IoU:0.37094\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 27 [960/1464 (66%)] ,Loss: 1.152203 ,Accuracy:84.95%, Train Mean IoU:0.36969\n",
      "Train Epoch: 27 [1120/1464 (77%)] ,Loss: 0.899088 ,Accuracy:85.09%, Train Mean IoU:0.36878\n",
      "Train Epoch: 27 [1280/1464 (87%)] ,Loss: 0.561034 ,Accuracy:85.02%, Train Mean IoU:0.36848\n",
      "Train Epoch: 27 [1440/1464 (98%)] ,Loss: 0.952492 ,Accuracy:84.95%, Train Mean IoU:0.36892\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 28 - - - - - - - -\n",
      "Train Epoch: 28 [160/1464 (11%)] ,Loss: 0.649826 ,Accuracy:86.42%, Train Mean IoU:0.36474\n",
      "Train Epoch: 28 [320/1464 (22%)] ,Loss: 1.327476 ,Accuracy:85.75%, Train Mean IoU:0.36953\n",
      "Train Epoch: 28 [480/1464 (33%)] ,Loss: 0.759570 ,Accuracy:85.43%, Train Mean IoU:0.37165\n",
      "Train Epoch: 28 [640/1464 (44%)] ,Loss: 0.743222 ,Accuracy:85.38%, Train Mean IoU:0.37024\n",
      "Train Epoch: 28 [800/1464 (55%)] ,Loss: 0.899942 ,Accuracy:85.12%, Train Mean IoU:0.36974\n",
      "Train Epoch: 28 [960/1464 (66%)] ,Loss: 0.932075 ,Accuracy:85.04%, Train Mean IoU:0.36940\n",
      "Train Epoch: 28 [1120/1464 (77%)] ,Loss: 0.614566 ,Accuracy:85.19%, Train Mean IoU:0.37036\n",
      "Train Epoch: 28 [1280/1464 (87%)] ,Loss: 0.792633 ,Accuracy:84.88%, Train Mean IoU:0.36894\n",
      "Train Epoch: 28 [1440/1464 (98%)] ,Loss: 0.708099 ,Accuracy:84.90%, Train Mean IoU:0.36868\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 29 - - - - - - - -\n",
      "Train Epoch: 29 [160/1464 (11%)] ,Loss: 0.816430 ,Accuracy:84.42%, Train Mean IoU:0.36236\n",
      "Train Epoch: 29 [320/1464 (22%)] ,Loss: 0.876532 ,Accuracy:85.25%, Train Mean IoU:0.37201\n",
      "Train Epoch: 29 [480/1464 (33%)] ,Loss: 0.926647 ,Accuracy:84.95%, Train Mean IoU:0.36897\n",
      "Train Epoch: 29 [640/1464 (44%)] ,Loss: 0.882713 ,Accuracy:84.64%, Train Mean IoU:0.36940\n",
      "Train Epoch: 29 [800/1464 (55%)] ,Loss: 0.539256 ,Accuracy:84.56%, Train Mean IoU:0.36581\n",
      "Train Epoch: 29 [960/1464 (66%)] ,Loss: 0.991252 ,Accuracy:84.87%, Train Mean IoU:0.36769\n",
      "Train Epoch: 29 [1120/1464 (77%)] ,Loss: 0.376074 ,Accuracy:84.97%, Train Mean IoU:0.36771\n",
      "Train Epoch: 29 [1280/1464 (87%)] ,Loss: 1.161459 ,Accuracy:84.65%, Train Mean IoU:0.36601\n",
      "Train Epoch: 29 [1440/1464 (98%)] ,Loss: 0.580238 ,Accuracy:84.88%, Train Mean IoU:0.36881\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 30 - - - - - - - -\n",
      "Train Epoch: 30 [160/1464 (11%)] ,Loss: 0.908408 ,Accuracy:85.11%, Train Mean IoU:0.37394\n",
      "Train Epoch: 30 [320/1464 (22%)] ,Loss: 0.706141 ,Accuracy:85.43%, Train Mean IoU:0.37276\n",
      "Train Epoch: 30 [480/1464 (33%)] ,Loss: 0.679978 ,Accuracy:85.27%, Train Mean IoU:0.37223\n",
      "Train Epoch: 30 [640/1464 (44%)] ,Loss: 1.091174 ,Accuracy:84.83%, Train Mean IoU:0.37082\n",
      "Train Epoch: 30 [800/1464 (55%)] ,Loss: 0.534868 ,Accuracy:84.88%, Train Mean IoU:0.37184\n",
      "Train Epoch: 30 [960/1464 (66%)] ,Loss: 0.643795 ,Accuracy:85.14%, Train Mean IoU:0.37179\n",
      "Train Epoch: 30 [1120/1464 (77%)] ,Loss: 0.678409 ,Accuracy:85.18%, Train Mean IoU:0.37294\n",
      "Train Epoch: 30 [1280/1464 (87%)] ,Loss: 0.546386 ,Accuracy:84.98%, Train Mean IoU:0.37051\n",
      "Train Epoch: 30 [1440/1464 (98%)] ,Loss: 0.931765 ,Accuracy:84.83%, Train Mean IoU:0.36839\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 31 - - - - - - - -\n",
      "LR is set to 0.0005\n",
      "Train Epoch: 31 [160/1464 (11%)] ,Loss: 0.955201 ,Accuracy:84.64%, Train Mean IoU:0.36774\n",
      "Train Epoch: 31 [320/1464 (22%)] ,Loss: 0.999946 ,Accuracy:84.66%, Train Mean IoU:0.36655\n",
      "Train Epoch: 31 [480/1464 (33%)] ,Loss: 0.933280 ,Accuracy:85.23%, Train Mean IoU:0.37201\n",
      "Train Epoch: 31 [640/1464 (44%)] ,Loss: 0.751511 ,Accuracy:85.06%, Train Mean IoU:0.37124\n",
      "Train Epoch: 31 [800/1464 (55%)] ,Loss: 0.745353 ,Accuracy:84.89%, Train Mean IoU:0.37201\n",
      "Train Epoch: 31 [960/1464 (66%)] ,Loss: 1.102495 ,Accuracy:84.84%, Train Mean IoU:0.36960\n",
      "Train Epoch: 31 [1120/1464 (77%)] ,Loss: 0.622888 ,Accuracy:84.75%, Train Mean IoU:0.36783\n",
      "Train Epoch: 31 [1280/1464 (87%)] ,Loss: 0.833418 ,Accuracy:84.79%, Train Mean IoU:0.36842\n",
      "Train Epoch: 31 [1440/1464 (98%)] ,Loss: 1.035147 ,Accuracy:84.91%, Train Mean IoU:0.36882\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 32 - - - - - - - -\n",
      "Train Epoch: 32 [160/1464 (11%)] ,Loss: 1.455591 ,Accuracy:83.24%, Train Mean IoU:0.36529\n",
      "Train Epoch: 32 [320/1464 (22%)] ,Loss: 0.698922 ,Accuracy:84.83%, Train Mean IoU:0.36823\n",
      "Train Epoch: 32 [480/1464 (33%)] ,Loss: 0.914083 ,Accuracy:85.25%, Train Mean IoU:0.37406\n",
      "Train Epoch: 32 [640/1464 (44%)] ,Loss: 0.891508 ,Accuracy:85.10%, Train Mean IoU:0.37039\n",
      "Train Epoch: 32 [800/1464 (55%)] ,Loss: 0.810798 ,Accuracy:85.02%, Train Mean IoU:0.36903\n",
      "Train Epoch: 32 [960/1464 (66%)] ,Loss: 0.693032 ,Accuracy:85.33%, Train Mean IoU:0.37000\n",
      "Train Epoch: 32 [1120/1464 (77%)] ,Loss: 0.828857 ,Accuracy:85.08%, Train Mean IoU:0.37053\n",
      "Train Epoch: 32 [1280/1464 (87%)] ,Loss: 0.818657 ,Accuracy:85.04%, Train Mean IoU:0.37105\n",
      "Train Epoch: 32 [1440/1464 (98%)] ,Loss: 1.247936 ,Accuracy:84.94%, Train Mean IoU:0.36875\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 33 - - - - - - - -\n",
      "Train Epoch: 33 [160/1464 (11%)] ,Loss: 1.000275 ,Accuracy:84.60%, Train Mean IoU:0.37108\n",
      "Train Epoch: 33 [320/1464 (22%)] ,Loss: 1.191899 ,Accuracy:84.43%, Train Mean IoU:0.36700\n",
      "Train Epoch: 33 [480/1464 (33%)] ,Loss: 0.896953 ,Accuracy:84.73%, Train Mean IoU:0.36920\n",
      "Train Epoch: 33 [640/1464 (44%)] ,Loss: 0.702083 ,Accuracy:84.95%, Train Mean IoU:0.36974\n",
      "Train Epoch: 33 [800/1464 (55%)] ,Loss: 0.823055 ,Accuracy:84.89%, Train Mean IoU:0.36862\n",
      "Train Epoch: 33 [960/1464 (66%)] ,Loss: 0.744412 ,Accuracy:85.11%, Train Mean IoU:0.37165\n",
      "Train Epoch: 33 [1120/1464 (77%)] ,Loss: 0.614011 ,Accuracy:85.18%, Train Mean IoU:0.37096\n",
      "Train Epoch: 33 [1280/1464 (87%)] ,Loss: 0.543099 ,Accuracy:84.97%, Train Mean IoU:0.37048\n",
      "Train Epoch: 33 [1440/1464 (98%)] ,Loss: 0.784905 ,Accuracy:84.96%, Train Mean IoU:0.36890\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 34 - - - - - - - -\n",
      "Train Epoch: 34 [160/1464 (11%)] ,Loss: 0.851268 ,Accuracy:86.01%, Train Mean IoU:0.36841\n",
      "Train Epoch: 34 [320/1464 (22%)] ,Loss: 0.702582 ,Accuracy:84.12%, Train Mean IoU:0.36343\n",
      "Train Epoch: 34 [480/1464 (33%)] ,Loss: 0.950324 ,Accuracy:84.44%, Train Mean IoU:0.36802\n",
      "Train Epoch: 34 [640/1464 (44%)] ,Loss: 0.882796 ,Accuracy:84.29%, Train Mean IoU:0.36717\n",
      "Train Epoch: 34 [800/1464 (55%)] ,Loss: 1.042444 ,Accuracy:84.74%, Train Mean IoU:0.36828\n",
      "Train Epoch: 34 [960/1464 (66%)] ,Loss: 0.734235 ,Accuracy:84.71%, Train Mean IoU:0.36811\n",
      "Train Epoch: 34 [1120/1464 (77%)] ,Loss: 0.661010 ,Accuracy:84.90%, Train Mean IoU:0.36953\n",
      "Train Epoch: 34 [1280/1464 (87%)] ,Loss: 0.528607 ,Accuracy:84.91%, Train Mean IoU:0.36909\n",
      "Train Epoch: 34 [1440/1464 (98%)] ,Loss: 0.730221 ,Accuracy:84.80%, Train Mean IoU:0.36867\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 35 - - - - - - - -\n",
      "Train Epoch: 35 [160/1464 (11%)] ,Loss: 0.818559 ,Accuracy:85.20%, Train Mean IoU:0.37548\n",
      "Train Epoch: 35 [320/1464 (22%)] ,Loss: 1.077293 ,Accuracy:83.55%, Train Mean IoU:0.36545\n",
      "Train Epoch: 35 [480/1464 (33%)] ,Loss: 1.076357 ,Accuracy:83.29%, Train Mean IoU:0.35927\n",
      "Train Epoch: 35 [640/1464 (44%)] ,Loss: 0.945483 ,Accuracy:83.89%, Train Mean IoU:0.36380\n",
      "Train Epoch: 35 [800/1464 (55%)] ,Loss: 0.563591 ,Accuracy:84.12%, Train Mean IoU:0.36418\n",
      "Train Epoch: 35 [960/1464 (66%)] ,Loss: 0.805521 ,Accuracy:84.40%, Train Mean IoU:0.36533\n",
      "Train Epoch: 35 [1120/1464 (77%)] ,Loss: 0.661469 ,Accuracy:84.60%, Train Mean IoU:0.36667\n",
      "Train Epoch: 35 [1280/1464 (87%)] ,Loss: 0.724264 ,Accuracy:84.69%, Train Mean IoU:0.36715\n",
      "Train Epoch: 35 [1440/1464 (98%)] ,Loss: 0.728803 ,Accuracy:84.83%, Train Mean IoU:0.36877\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 36 - - - - - - - -\n",
      "Train Epoch: 36 [160/1464 (11%)] ,Loss: 0.932446 ,Accuracy:86.02%, Train Mean IoU:0.37531\n",
      "Train Epoch: 36 [320/1464 (22%)] ,Loss: 0.728492 ,Accuracy:85.97%, Train Mean IoU:0.37669\n",
      "Train Epoch: 36 [480/1464 (33%)] ,Loss: 0.766060 ,Accuracy:85.72%, Train Mean IoU:0.37466\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 36 [640/1464 (44%)] ,Loss: 0.855983 ,Accuracy:85.01%, Train Mean IoU:0.37078\n",
      "Train Epoch: 36 [800/1464 (55%)] ,Loss: 0.822277 ,Accuracy:85.06%, Train Mean IoU:0.36814\n",
      "Train Epoch: 36 [960/1464 (66%)] ,Loss: 1.133113 ,Accuracy:84.79%, Train Mean IoU:0.36613\n",
      "Train Epoch: 36 [1120/1464 (77%)] ,Loss: 0.820957 ,Accuracy:84.79%, Train Mean IoU:0.36646\n",
      "Train Epoch: 36 [1280/1464 (87%)] ,Loss: 0.864756 ,Accuracy:84.65%, Train Mean IoU:0.36625\n",
      "Train Epoch: 36 [1440/1464 (98%)] ,Loss: 1.002903 ,Accuracy:84.83%, Train Mean IoU:0.36743\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 37 - - - - - - - -\n",
      "Train Epoch: 37 [160/1464 (11%)] ,Loss: 0.783950 ,Accuracy:86.31%, Train Mean IoU:0.38502\n",
      "Train Epoch: 37 [320/1464 (22%)] ,Loss: 0.644767 ,Accuracy:85.13%, Train Mean IoU:0.37131\n",
      "Train Epoch: 37 [480/1464 (33%)] ,Loss: 0.715990 ,Accuracy:85.28%, Train Mean IoU:0.37014\n",
      "Train Epoch: 37 [640/1464 (44%)] ,Loss: 0.880177 ,Accuracy:84.79%, Train Mean IoU:0.36799\n",
      "Train Epoch: 37 [800/1464 (55%)] ,Loss: 1.092230 ,Accuracy:84.43%, Train Mean IoU:0.36464\n",
      "Train Epoch: 37 [960/1464 (66%)] ,Loss: 0.638085 ,Accuracy:84.65%, Train Mean IoU:0.36610\n",
      "Train Epoch: 37 [1120/1464 (77%)] ,Loss: 0.474332 ,Accuracy:84.79%, Train Mean IoU:0.36745\n",
      "Train Epoch: 37 [1280/1464 (87%)] ,Loss: 0.687512 ,Accuracy:85.01%, Train Mean IoU:0.36864\n",
      "Train Epoch: 37 [1440/1464 (98%)] ,Loss: 0.975278 ,Accuracy:84.95%, Train Mean IoU:0.36887\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 38 - - - - - - - -\n",
      "Train Epoch: 38 [160/1464 (11%)] ,Loss: 0.657243 ,Accuracy:85.80%, Train Mean IoU:0.36620\n",
      "Train Epoch: 38 [320/1464 (22%)] ,Loss: 0.797639 ,Accuracy:85.33%, Train Mean IoU:0.36808\n",
      "Train Epoch: 38 [480/1464 (33%)] ,Loss: 0.503240 ,Accuracy:85.44%, Train Mean IoU:0.36925\n",
      "Train Epoch: 38 [640/1464 (44%)] ,Loss: 0.863878 ,Accuracy:85.09%, Train Mean IoU:0.36894\n",
      "Train Epoch: 38 [800/1464 (55%)] ,Loss: 0.470693 ,Accuracy:84.95%, Train Mean IoU:0.36703\n",
      "Train Epoch: 38 [960/1464 (66%)] ,Loss: 0.730422 ,Accuracy:84.91%, Train Mean IoU:0.36792\n",
      "Train Epoch: 38 [1120/1464 (77%)] ,Loss: 1.117258 ,Accuracy:84.84%, Train Mean IoU:0.36909\n",
      "Train Epoch: 38 [1280/1464 (87%)] ,Loss: 0.999591 ,Accuracy:84.67%, Train Mean IoU:0.36729\n",
      "Train Epoch: 38 [1440/1464 (98%)] ,Loss: 0.582493 ,Accuracy:84.93%, Train Mean IoU:0.36835\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 39 - - - - - - - -\n",
      "Train Epoch: 39 [160/1464 (11%)] ,Loss: 0.671095 ,Accuracy:85.57%, Train Mean IoU:0.37428\n",
      "Train Epoch: 39 [320/1464 (22%)] ,Loss: 0.763667 ,Accuracy:84.64%, Train Mean IoU:0.36582\n",
      "Train Epoch: 39 [480/1464 (33%)] ,Loss: 0.712642 ,Accuracy:84.54%, Train Mean IoU:0.36807\n",
      "Train Epoch: 39 [640/1464 (44%)] ,Loss: 1.257475 ,Accuracy:84.90%, Train Mean IoU:0.36956\n",
      "Train Epoch: 39 [800/1464 (55%)] ,Loss: 0.819202 ,Accuracy:84.97%, Train Mean IoU:0.37030\n",
      "Train Epoch: 39 [960/1464 (66%)] ,Loss: 0.558702 ,Accuracy:85.23%, Train Mean IoU:0.36963\n",
      "Train Epoch: 39 [1120/1464 (77%)] ,Loss: 0.646396 ,Accuracy:84.92%, Train Mean IoU:0.36969\n",
      "Train Epoch: 39 [1280/1464 (87%)] ,Loss: 0.809873 ,Accuracy:85.07%, Train Mean IoU:0.36900\n",
      "Train Epoch: 39 [1440/1464 (98%)] ,Loss: 1.062162 ,Accuracy:84.96%, Train Mean IoU:0.36887\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 40 - - - - - - - -\n",
      "Train Epoch: 40 [160/1464 (11%)] ,Loss: 0.842121 ,Accuracy:83.14%, Train Mean IoU:0.35774\n",
      "Train Epoch: 40 [320/1464 (22%)] ,Loss: 0.673212 ,Accuracy:83.45%, Train Mean IoU:0.35871\n",
      "Train Epoch: 40 [480/1464 (33%)] ,Loss: 1.001909 ,Accuracy:83.70%, Train Mean IoU:0.36219\n",
      "Train Epoch: 40 [640/1464 (44%)] ,Loss: 0.880149 ,Accuracy:84.15%, Train Mean IoU:0.36507\n",
      "Train Epoch: 40 [800/1464 (55%)] ,Loss: 0.732074 ,Accuracy:84.63%, Train Mean IoU:0.36672\n",
      "Train Epoch: 40 [960/1464 (66%)] ,Loss: 0.783796 ,Accuracy:84.65%, Train Mean IoU:0.36647\n",
      "Train Epoch: 40 [1120/1464 (77%)] ,Loss: 0.912610 ,Accuracy:84.85%, Train Mean IoU:0.36742\n",
      "Train Epoch: 40 [1280/1464 (87%)] ,Loss: 0.738508 ,Accuracy:84.99%, Train Mean IoU:0.36827\n",
      "Train Epoch: 40 [1440/1464 (98%)] ,Loss: 0.872154 ,Accuracy:84.87%, Train Mean IoU:0.36845\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 41 - - - - - - - -\n",
      "Train Epoch: 41 [160/1464 (11%)] ,Loss: 1.065593 ,Accuracy:85.15%, Train Mean IoU:0.37048\n",
      "Train Epoch: 41 [320/1464 (22%)] ,Loss: 0.824039 ,Accuracy:85.69%, Train Mean IoU:0.36795\n",
      "Train Epoch: 41 [480/1464 (33%)] ,Loss: 0.618973 ,Accuracy:85.44%, Train Mean IoU:0.36743\n",
      "Train Epoch: 41 [640/1464 (44%)] ,Loss: 1.026550 ,Accuracy:84.52%, Train Mean IoU:0.36540\n",
      "Train Epoch: 41 [800/1464 (55%)] ,Loss: 0.883684 ,Accuracy:84.89%, Train Mean IoU:0.37026\n",
      "Train Epoch: 41 [960/1464 (66%)] ,Loss: 0.571919 ,Accuracy:84.93%, Train Mean IoU:0.36969\n",
      "Train Epoch: 41 [1120/1464 (77%)] ,Loss: 0.965091 ,Accuracy:84.74%, Train Mean IoU:0.36887\n",
      "Train Epoch: 41 [1280/1464 (87%)] ,Loss: 0.732933 ,Accuracy:84.90%, Train Mean IoU:0.36931\n",
      "Train Epoch: 41 [1440/1464 (98%)] ,Loss: 0.487096 ,Accuracy:84.89%, Train Mean IoU:0.36906\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 42 - - - - - - - -\n",
      "Train Epoch: 42 [160/1464 (11%)] ,Loss: 0.848013 ,Accuracy:84.59%, Train Mean IoU:0.37001\n",
      "Train Epoch: 42 [320/1464 (22%)] ,Loss: 1.060763 ,Accuracy:84.69%, Train Mean IoU:0.37108\n",
      "Train Epoch: 42 [480/1464 (33%)] ,Loss: 0.828700 ,Accuracy:84.60%, Train Mean IoU:0.36884\n",
      "Train Epoch: 42 [640/1464 (44%)] ,Loss: 0.935564 ,Accuracy:84.88%, Train Mean IoU:0.36962\n",
      "Train Epoch: 42 [800/1464 (55%)] ,Loss: 0.879978 ,Accuracy:84.91%, Train Mean IoU:0.37055\n",
      "Train Epoch: 42 [960/1464 (66%)] ,Loss: 0.989045 ,Accuracy:84.82%, Train Mean IoU:0.37053\n",
      "Train Epoch: 42 [1120/1464 (77%)] ,Loss: 0.937242 ,Accuracy:84.84%, Train Mean IoU:0.36978\n",
      "Train Epoch: 42 [1280/1464 (87%)] ,Loss: 0.648801 ,Accuracy:84.82%, Train Mean IoU:0.36834\n",
      "Train Epoch: 42 [1440/1464 (98%)] ,Loss: 1.074446 ,Accuracy:84.96%, Train Mean IoU:0.36893\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 43 - - - - - - - -\n",
      "Train Epoch: 43 [160/1464 (11%)] ,Loss: 1.180359 ,Accuracy:84.68%, Train Mean IoU:0.36808\n",
      "Train Epoch: 43 [320/1464 (22%)] ,Loss: 1.089532 ,Accuracy:85.19%, Train Mean IoU:0.36773\n",
      "Train Epoch: 43 [480/1464 (33%)] ,Loss: 0.480166 ,Accuracy:85.56%, Train Mean IoU:0.37406\n",
      "Train Epoch: 43 [640/1464 (44%)] ,Loss: 0.727752 ,Accuracy:84.93%, Train Mean IoU:0.36897\n",
      "Train Epoch: 43 [800/1464 (55%)] ,Loss: 0.672112 ,Accuracy:84.98%, Train Mean IoU:0.36903\n",
      "Train Epoch: 43 [960/1464 (66%)] ,Loss: 0.710721 ,Accuracy:85.05%, Train Mean IoU:0.36930\n",
      "Train Epoch: 43 [1120/1464 (77%)] ,Loss: 0.926764 ,Accuracy:84.81%, Train Mean IoU:0.36777\n",
      "Train Epoch: 43 [1280/1464 (87%)] ,Loss: 1.016857 ,Accuracy:84.98%, Train Mean IoU:0.36878\n",
      "Train Epoch: 43 [1440/1464 (98%)] ,Loss: 0.808284 ,Accuracy:84.89%, Train Mean IoU:0.36821\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 44 - - - - - - - -\n",
      "Train Epoch: 44 [160/1464 (11%)] ,Loss: 0.599312 ,Accuracy:84.72%, Train Mean IoU:0.36345\n",
      "Train Epoch: 44 [320/1464 (22%)] ,Loss: 0.652556 ,Accuracy:84.93%, Train Mean IoU:0.36504\n",
      "Train Epoch: 44 [480/1464 (33%)] ,Loss: 1.089948 ,Accuracy:85.09%, Train Mean IoU:0.36471\n",
      "Train Epoch: 44 [640/1464 (44%)] ,Loss: 0.666027 ,Accuracy:85.08%, Train Mean IoU:0.36847\n",
      "Train Epoch: 44 [800/1464 (55%)] ,Loss: 0.771444 ,Accuracy:85.00%, Train Mean IoU:0.36715\n",
      "Train Epoch: 44 [960/1464 (66%)] ,Loss: 0.625810 ,Accuracy:84.93%, Train Mean IoU:0.36812\n",
      "Train Epoch: 44 [1120/1464 (77%)] ,Loss: 0.913142 ,Accuracy:84.87%, Train Mean IoU:0.36940\n",
      "Train Epoch: 44 [1280/1464 (87%)] ,Loss: 1.026305 ,Accuracy:84.81%, Train Mean IoU:0.36813\n",
      "Train Epoch: 44 [1440/1464 (98%)] ,Loss: 0.735304 ,Accuracy:84.85%, Train Mean IoU:0.36804\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 45 - - - - - - - -\n",
      "Train Epoch: 45 [160/1464 (11%)] ,Loss: 0.796311 ,Accuracy:85.62%, Train Mean IoU:0.36470\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 45 [320/1464 (22%)] ,Loss: 0.591590 ,Accuracy:84.27%, Train Mean IoU:0.36981\n",
      "Train Epoch: 45 [480/1464 (33%)] ,Loss: 0.571124 ,Accuracy:84.74%, Train Mean IoU:0.37448\n",
      "Train Epoch: 45 [640/1464 (44%)] ,Loss: 0.484832 ,Accuracy:84.53%, Train Mean IoU:0.37286\n",
      "Train Epoch: 45 [800/1464 (55%)] ,Loss: 0.640872 ,Accuracy:84.82%, Train Mean IoU:0.37398\n",
      "Train Epoch: 45 [960/1464 (66%)] ,Loss: 0.931313 ,Accuracy:84.74%, Train Mean IoU:0.37205\n",
      "Train Epoch: 45 [1120/1464 (77%)] ,Loss: 0.817692 ,Accuracy:84.64%, Train Mean IoU:0.36928\n",
      "Train Epoch: 45 [1280/1464 (87%)] ,Loss: 0.717358 ,Accuracy:84.68%, Train Mean IoU:0.36775\n",
      "Train Epoch: 45 [1440/1464 (98%)] ,Loss: 0.613119 ,Accuracy:84.88%, Train Mean IoU:0.36862\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 46 - - - - - - - -\n",
      "Train Epoch: 46 [160/1464 (11%)] ,Loss: 0.510176 ,Accuracy:84.22%, Train Mean IoU:0.36795\n",
      "Train Epoch: 46 [320/1464 (22%)] ,Loss: 1.100514 ,Accuracy:84.96%, Train Mean IoU:0.36317\n",
      "Train Epoch: 46 [480/1464 (33%)] ,Loss: 1.317401 ,Accuracy:84.65%, Train Mean IoU:0.36332\n",
      "Train Epoch: 46 [640/1464 (44%)] ,Loss: 0.907131 ,Accuracy:84.89%, Train Mean IoU:0.36555\n",
      "Train Epoch: 46 [800/1464 (55%)] ,Loss: 0.595597 ,Accuracy:84.43%, Train Mean IoU:0.36453\n",
      "Train Epoch: 46 [960/1464 (66%)] ,Loss: 0.598907 ,Accuracy:84.55%, Train Mean IoU:0.36603\n",
      "Train Epoch: 46 [1120/1464 (77%)] ,Loss: 0.857617 ,Accuracy:84.93%, Train Mean IoU:0.36720\n",
      "Train Epoch: 46 [1280/1464 (87%)] ,Loss: 0.885632 ,Accuracy:84.88%, Train Mean IoU:0.36774\n",
      "Train Epoch: 46 [1440/1464 (98%)] ,Loss: 0.691821 ,Accuracy:84.91%, Train Mean IoU:0.36836\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 47 - - - - - - - -\n",
      "Train Epoch: 47 [160/1464 (11%)] ,Loss: 0.674129 ,Accuracy:83.83%, Train Mean IoU:0.35510\n",
      "Train Epoch: 47 [320/1464 (22%)] ,Loss: 0.734587 ,Accuracy:83.44%, Train Mean IoU:0.36295\n",
      "Train Epoch: 47 [480/1464 (33%)] ,Loss: 1.092736 ,Accuracy:84.18%, Train Mean IoU:0.36714\n",
      "Train Epoch: 47 [640/1464 (44%)] ,Loss: 0.857875 ,Accuracy:84.35%, Train Mean IoU:0.37064\n",
      "Train Epoch: 47 [800/1464 (55%)] ,Loss: 0.771808 ,Accuracy:84.35%, Train Mean IoU:0.36892\n",
      "Train Epoch: 47 [960/1464 (66%)] ,Loss: 0.531953 ,Accuracy:84.66%, Train Mean IoU:0.36956\n",
      "Train Epoch: 47 [1120/1464 (77%)] ,Loss: 0.571645 ,Accuracy:84.84%, Train Mean IoU:0.36991\n",
      "Train Epoch: 47 [1280/1464 (87%)] ,Loss: 0.736383 ,Accuracy:84.84%, Train Mean IoU:0.36895\n",
      "Train Epoch: 47 [1440/1464 (98%)] ,Loss: 1.043742 ,Accuracy:84.91%, Train Mean IoU:0.36869\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 48 - - - - - - - -\n",
      "Train Epoch: 48 [160/1464 (11%)] ,Loss: 0.572025 ,Accuracy:84.72%, Train Mean IoU:0.36799\n",
      "Train Epoch: 48 [320/1464 (22%)] ,Loss: 0.498378 ,Accuracy:85.13%, Train Mean IoU:0.37466\n",
      "Train Epoch: 48 [480/1464 (33%)] ,Loss: 0.948053 ,Accuracy:84.76%, Train Mean IoU:0.37163\n",
      "Train Epoch: 48 [640/1464 (44%)] ,Loss: 0.739361 ,Accuracy:84.85%, Train Mean IoU:0.36902\n",
      "Train Epoch: 48 [800/1464 (55%)] ,Loss: 1.278468 ,Accuracy:84.52%, Train Mean IoU:0.36766\n",
      "Train Epoch: 48 [960/1464 (66%)] ,Loss: 0.558155 ,Accuracy:84.66%, Train Mean IoU:0.36799\n",
      "Train Epoch: 48 [1120/1464 (77%)] ,Loss: 1.091264 ,Accuracy:84.56%, Train Mean IoU:0.36721\n",
      "Train Epoch: 48 [1280/1464 (87%)] ,Loss: 0.882243 ,Accuracy:84.58%, Train Mean IoU:0.36662\n",
      "Train Epoch: 48 [1440/1464 (98%)] ,Loss: 1.047977 ,Accuracy:84.82%, Train Mean IoU:0.36813\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 49 - - - - - - - -\n",
      "Train Epoch: 49 [160/1464 (11%)] ,Loss: 0.681044 ,Accuracy:85.73%, Train Mean IoU:0.37427\n",
      "Train Epoch: 49 [320/1464 (22%)] ,Loss: 1.218517 ,Accuracy:84.91%, Train Mean IoU:0.36630\n",
      "Train Epoch: 49 [480/1464 (33%)] ,Loss: 0.986632 ,Accuracy:84.99%, Train Mean IoU:0.36653\n",
      "Train Epoch: 49 [640/1464 (44%)] ,Loss: 0.961714 ,Accuracy:84.91%, Train Mean IoU:0.36778\n",
      "Train Epoch: 49 [800/1464 (55%)] ,Loss: 0.678629 ,Accuracy:84.94%, Train Mean IoU:0.36839\n",
      "Train Epoch: 49 [960/1464 (66%)] ,Loss: 1.089872 ,Accuracy:84.88%, Train Mean IoU:0.36849\n",
      "Train Epoch: 49 [1120/1464 (77%)] ,Loss: 1.133031 ,Accuracy:84.87%, Train Mean IoU:0.36840\n",
      "Train Epoch: 49 [1280/1464 (87%)] ,Loss: 0.646297 ,Accuracy:85.12%, Train Mean IoU:0.36979\n",
      "Train Epoch: 49 [1440/1464 (98%)] ,Loss: 1.138971 ,Accuracy:84.93%, Train Mean IoU:0.36855\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 50 - - - - - - - -\n",
      "Train Epoch: 50 [160/1464 (11%)] ,Loss: 1.017617 ,Accuracy:85.02%, Train Mean IoU:0.36765\n",
      "Train Epoch: 50 [320/1464 (22%)] ,Loss: 0.867762 ,Accuracy:85.23%, Train Mean IoU:0.36891\n",
      "Train Epoch: 50 [480/1464 (33%)] ,Loss: 0.688114 ,Accuracy:85.13%, Train Mean IoU:0.36817\n",
      "Train Epoch: 50 [640/1464 (44%)] ,Loss: 0.841007 ,Accuracy:84.69%, Train Mean IoU:0.36749\n",
      "Train Epoch: 50 [800/1464 (55%)] ,Loss: 0.765463 ,Accuracy:84.83%, Train Mean IoU:0.36728\n",
      "Train Epoch: 50 [960/1464 (66%)] ,Loss: 0.698295 ,Accuracy:85.07%, Train Mean IoU:0.36974\n",
      "Train Epoch: 50 [1120/1464 (77%)] ,Loss: 0.535888 ,Accuracy:85.11%, Train Mean IoU:0.36900\n",
      "Train Epoch: 50 [1280/1464 (87%)] ,Loss: 0.852927 ,Accuracy:85.08%, Train Mean IoU:0.36987\n",
      "Train Epoch: 50 [1440/1464 (98%)] ,Loss: 0.910613 ,Accuracy:85.00%, Train Mean IoU:0.36936\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 51 - - - - - - - -\n",
      "Train Epoch: 51 [160/1464 (11%)] ,Loss: 0.699210 ,Accuracy:86.09%, Train Mean IoU:0.38026\n",
      "Train Epoch: 51 [320/1464 (22%)] ,Loss: 0.781273 ,Accuracy:85.56%, Train Mean IoU:0.37676\n",
      "Train Epoch: 51 [480/1464 (33%)] ,Loss: 1.056278 ,Accuracy:84.59%, Train Mean IoU:0.37164\n",
      "Train Epoch: 51 [640/1464 (44%)] ,Loss: 1.329440 ,Accuracy:84.41%, Train Mean IoU:0.36980\n",
      "Train Epoch: 51 [800/1464 (55%)] ,Loss: 0.689385 ,Accuracy:84.78%, Train Mean IoU:0.36972\n",
      "Train Epoch: 51 [960/1464 (66%)] ,Loss: 0.884962 ,Accuracy:84.88%, Train Mean IoU:0.37011\n",
      "Train Epoch: 51 [1120/1464 (77%)] ,Loss: 0.715243 ,Accuracy:84.96%, Train Mean IoU:0.37065\n",
      "Train Epoch: 51 [1280/1464 (87%)] ,Loss: 0.973605 ,Accuracy:84.85%, Train Mean IoU:0.36932\n",
      "Train Epoch: 51 [1440/1464 (98%)] ,Loss: 0.965008 ,Accuracy:84.88%, Train Mean IoU:0.36866\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 52 - - - - - - - -\n",
      "Train Epoch: 52 [160/1464 (11%)] ,Loss: 1.059886 ,Accuracy:84.63%, Train Mean IoU:0.35581\n",
      "Train Epoch: 52 [320/1464 (22%)] ,Loss: 0.946246 ,Accuracy:84.91%, Train Mean IoU:0.36308\n",
      "Train Epoch: 52 [480/1464 (33%)] ,Loss: 0.638134 ,Accuracy:84.94%, Train Mean IoU:0.36456\n",
      "Train Epoch: 52 [640/1464 (44%)] ,Loss: 0.662947 ,Accuracy:84.76%, Train Mean IoU:0.36601\n",
      "Train Epoch: 52 [800/1464 (55%)] ,Loss: 0.612514 ,Accuracy:84.72%, Train Mean IoU:0.36705\n",
      "Train Epoch: 52 [960/1464 (66%)] ,Loss: 0.623054 ,Accuracy:84.95%, Train Mean IoU:0.36758\n",
      "Train Epoch: 52 [1120/1464 (77%)] ,Loss: 0.768590 ,Accuracy:84.63%, Train Mean IoU:0.36590\n",
      "Train Epoch: 52 [1280/1464 (87%)] ,Loss: 0.823553 ,Accuracy:84.87%, Train Mean IoU:0.36950\n",
      "Train Epoch: 52 [1440/1464 (98%)] ,Loss: 1.008659 ,Accuracy:84.88%, Train Mean IoU:0.36904\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 53 - - - - - - - -\n",
      "Train Epoch: 53 [160/1464 (11%)] ,Loss: 0.769098 ,Accuracy:86.27%, Train Mean IoU:0.36802\n",
      "Train Epoch: 53 [320/1464 (22%)] ,Loss: 0.542683 ,Accuracy:86.19%, Train Mean IoU:0.37295\n",
      "Train Epoch: 53 [480/1464 (33%)] ,Loss: 0.662605 ,Accuracy:85.39%, Train Mean IoU:0.37083\n",
      "Train Epoch: 53 [640/1464 (44%)] ,Loss: 0.966423 ,Accuracy:85.13%, Train Mean IoU:0.37178\n",
      "Train Epoch: 53 [800/1464 (55%)] ,Loss: 0.629540 ,Accuracy:85.21%, Train Mean IoU:0.37082\n",
      "Train Epoch: 53 [960/1464 (66%)] ,Loss: 0.659003 ,Accuracy:85.21%, Train Mean IoU:0.36986\n",
      "Train Epoch: 53 [1120/1464 (77%)] ,Loss: 0.779136 ,Accuracy:85.09%, Train Mean IoU:0.36933\n",
      "Train Epoch: 53 [1280/1464 (87%)] ,Loss: 0.910918 ,Accuracy:84.97%, Train Mean IoU:0.36925\n",
      "Train Epoch: 53 [1440/1464 (98%)] ,Loss: 0.789611 ,Accuracy:84.83%, Train Mean IoU:0.36815\n",
      "- - - - - test stage- - - - - - - -\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 54 - - - - - - - -\n",
      "Train Epoch: 54 [160/1464 (11%)] ,Loss: 0.937330 ,Accuracy:84.95%, Train Mean IoU:0.36198\n",
      "Train Epoch: 54 [320/1464 (22%)] ,Loss: 0.960804 ,Accuracy:84.85%, Train Mean IoU:0.36304\n",
      "Train Epoch: 54 [480/1464 (33%)] ,Loss: 0.725337 ,Accuracy:85.01%, Train Mean IoU:0.36521\n",
      "Train Epoch: 54 [640/1464 (44%)] ,Loss: 0.795570 ,Accuracy:85.49%, Train Mean IoU:0.37101\n",
      "Train Epoch: 54 [800/1464 (55%)] ,Loss: 0.575039 ,Accuracy:85.32%, Train Mean IoU:0.37088\n",
      "Train Epoch: 54 [960/1464 (66%)] ,Loss: 1.034183 ,Accuracy:85.22%, Train Mean IoU:0.36990\n",
      "Train Epoch: 54 [1120/1464 (77%)] ,Loss: 0.645227 ,Accuracy:85.17%, Train Mean IoU:0.37016\n",
      "Train Epoch: 54 [1280/1464 (87%)] ,Loss: 0.712551 ,Accuracy:85.01%, Train Mean IoU:0.36809\n",
      "Train Epoch: 54 [1440/1464 (98%)] ,Loss: 0.789541 ,Accuracy:84.86%, Train Mean IoU:0.36853\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 55 - - - - - - - -\n",
      "Train Epoch: 55 [160/1464 (11%)] ,Loss: 0.792525 ,Accuracy:85.51%, Train Mean IoU:0.36670\n",
      "Train Epoch: 55 [320/1464 (22%)] ,Loss: 0.867439 ,Accuracy:85.50%, Train Mean IoU:0.36918\n",
      "Train Epoch: 55 [480/1464 (33%)] ,Loss: 0.880886 ,Accuracy:85.45%, Train Mean IoU:0.37476\n",
      "Train Epoch: 55 [640/1464 (44%)] ,Loss: 0.991008 ,Accuracy:84.82%, Train Mean IoU:0.36987\n",
      "Train Epoch: 55 [800/1464 (55%)] ,Loss: 0.578429 ,Accuracy:84.87%, Train Mean IoU:0.36811\n",
      "Train Epoch: 55 [960/1464 (66%)] ,Loss: 1.052602 ,Accuracy:84.81%, Train Mean IoU:0.36763\n",
      "Train Epoch: 55 [1120/1464 (77%)] ,Loss: 0.916918 ,Accuracy:84.67%, Train Mean IoU:0.36637\n",
      "Train Epoch: 55 [1280/1464 (87%)] ,Loss: 0.983770 ,Accuracy:84.84%, Train Mean IoU:0.36708\n",
      "Train Epoch: 55 [1440/1464 (98%)] ,Loss: 1.109875 ,Accuracy:84.88%, Train Mean IoU:0.36825\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 56 - - - - - - - -\n",
      "Train Epoch: 56 [160/1464 (11%)] ,Loss: 0.601396 ,Accuracy:85.90%, Train Mean IoU:0.38693\n",
      "Train Epoch: 56 [320/1464 (22%)] ,Loss: 0.702800 ,Accuracy:86.03%, Train Mean IoU:0.37218\n",
      "Train Epoch: 56 [480/1464 (33%)] ,Loss: 0.678327 ,Accuracy:85.33%, Train Mean IoU:0.37070\n",
      "Train Epoch: 56 [640/1464 (44%)] ,Loss: 0.834160 ,Accuracy:84.81%, Train Mean IoU:0.37004\n",
      "Train Epoch: 56 [800/1464 (55%)] ,Loss: 1.074792 ,Accuracy:84.77%, Train Mean IoU:0.36884\n",
      "Train Epoch: 56 [960/1464 (66%)] ,Loss: 0.836165 ,Accuracy:85.00%, Train Mean IoU:0.36924\n",
      "Train Epoch: 56 [1120/1464 (77%)] ,Loss: 0.919386 ,Accuracy:84.77%, Train Mean IoU:0.36736\n",
      "Train Epoch: 56 [1280/1464 (87%)] ,Loss: 0.451587 ,Accuracy:84.92%, Train Mean IoU:0.36867\n",
      "Train Epoch: 56 [1440/1464 (98%)] ,Loss: 0.851559 ,Accuracy:84.86%, Train Mean IoU:0.36838\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 57 - - - - - - - -\n",
      "Train Epoch: 57 [160/1464 (11%)] ,Loss: 1.041424 ,Accuracy:84.40%, Train Mean IoU:0.36530\n",
      "Train Epoch: 57 [320/1464 (22%)] ,Loss: 1.088870 ,Accuracy:85.29%, Train Mean IoU:0.37125\n",
      "Train Epoch: 57 [480/1464 (33%)] ,Loss: 0.642237 ,Accuracy:85.22%, Train Mean IoU:0.36928\n",
      "Train Epoch: 57 [640/1464 (44%)] ,Loss: 0.785463 ,Accuracy:84.98%, Train Mean IoU:0.36881\n",
      "Train Epoch: 57 [800/1464 (55%)] ,Loss: 0.829923 ,Accuracy:85.32%, Train Mean IoU:0.37157\n",
      "Train Epoch: 57 [960/1464 (66%)] ,Loss: 0.544690 ,Accuracy:85.33%, Train Mean IoU:0.37068\n",
      "Train Epoch: 57 [1120/1464 (77%)] ,Loss: 0.891834 ,Accuracy:85.21%, Train Mean IoU:0.36959\n",
      "Train Epoch: 57 [1280/1464 (87%)] ,Loss: 1.069352 ,Accuracy:84.98%, Train Mean IoU:0.36898\n",
      "Train Epoch: 57 [1440/1464 (98%)] ,Loss: 0.914279 ,Accuracy:84.87%, Train Mean IoU:0.36899\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 58 - - - - - - - -\n",
      "Train Epoch: 58 [160/1464 (11%)] ,Loss: 1.005209 ,Accuracy:84.13%, Train Mean IoU:0.37019\n",
      "Train Epoch: 58 [320/1464 (22%)] ,Loss: 0.688320 ,Accuracy:84.69%, Train Mean IoU:0.36847\n",
      "Train Epoch: 58 [480/1464 (33%)] ,Loss: 0.924767 ,Accuracy:84.12%, Train Mean IoU:0.36503\n",
      "Train Epoch: 58 [640/1464 (44%)] ,Loss: 0.697661 ,Accuracy:84.84%, Train Mean IoU:0.36599\n",
      "Train Epoch: 58 [800/1464 (55%)] ,Loss: 0.775363 ,Accuracy:84.81%, Train Mean IoU:0.36709\n",
      "Train Epoch: 58 [960/1464 (66%)] ,Loss: 0.725155 ,Accuracy:84.71%, Train Mean IoU:0.36663\n",
      "Train Epoch: 58 [1120/1464 (77%)] ,Loss: 0.898361 ,Accuracy:84.62%, Train Mean IoU:0.36654\n",
      "Train Epoch: 58 [1280/1464 (87%)] ,Loss: 0.929832 ,Accuracy:84.78%, Train Mean IoU:0.36806\n",
      "Train Epoch: 58 [1440/1464 (98%)] ,Loss: 0.574022 ,Accuracy:84.92%, Train Mean IoU:0.36862\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 59 - - - - - - - -\n",
      "Train Epoch: 59 [160/1464 (11%)] ,Loss: 1.034743 ,Accuracy:84.98%, Train Mean IoU:0.36717\n",
      "Train Epoch: 59 [320/1464 (22%)] ,Loss: 0.763588 ,Accuracy:84.55%, Train Mean IoU:0.36274\n",
      "Train Epoch: 59 [480/1464 (33%)] ,Loss: 0.919140 ,Accuracy:84.49%, Train Mean IoU:0.36530\n",
      "Train Epoch: 59 [640/1464 (44%)] ,Loss: 0.837725 ,Accuracy:84.41%, Train Mean IoU:0.36821\n",
      "Train Epoch: 59 [800/1464 (55%)] ,Loss: 0.644390 ,Accuracy:84.68%, Train Mean IoU:0.36878\n",
      "Train Epoch: 59 [960/1464 (66%)] ,Loss: 0.540188 ,Accuracy:84.94%, Train Mean IoU:0.36959\n",
      "Train Epoch: 59 [1120/1464 (77%)] ,Loss: 0.851735 ,Accuracy:85.07%, Train Mean IoU:0.37076\n",
      "Train Epoch: 59 [1280/1464 (87%)] ,Loss: 0.793543 ,Accuracy:85.08%, Train Mean IoU:0.36946\n",
      "Train Epoch: 59 [1440/1464 (98%)] ,Loss: 0.761739 ,Accuracy:84.96%, Train Mean IoU:0.36897\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 60 - - - - - - - -\n",
      "Train Epoch: 60 [160/1464 (11%)] ,Loss: 1.040478 ,Accuracy:83.34%, Train Mean IoU:0.35366\n",
      "Train Epoch: 60 [320/1464 (22%)] ,Loss: 0.984739 ,Accuracy:84.80%, Train Mean IoU:0.36475\n",
      "Train Epoch: 60 [480/1464 (33%)] ,Loss: 0.972455 ,Accuracy:85.31%, Train Mean IoU:0.36733\n",
      "Train Epoch: 60 [640/1464 (44%)] ,Loss: 0.807732 ,Accuracy:85.13%, Train Mean IoU:0.36664\n",
      "Train Epoch: 60 [800/1464 (55%)] ,Loss: 0.626945 ,Accuracy:85.20%, Train Mean IoU:0.36852\n",
      "Train Epoch: 60 [960/1464 (66%)] ,Loss: 1.192582 ,Accuracy:85.13%, Train Mean IoU:0.36911\n",
      "Train Epoch: 60 [1120/1464 (77%)] ,Loss: 0.660625 ,Accuracy:85.17%, Train Mean IoU:0.36988\n",
      "Train Epoch: 60 [1280/1464 (87%)] ,Loss: 0.668012 ,Accuracy:84.90%, Train Mean IoU:0.36944\n",
      "Train Epoch: 60 [1440/1464 (98%)] ,Loss: 1.050232 ,Accuracy:84.87%, Train Mean IoU:0.36866\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 61 - - - - - - - -\n",
      "LR is set to 0.00025\n",
      "Train Epoch: 61 [160/1464 (11%)] ,Loss: 0.861598 ,Accuracy:84.67%, Train Mean IoU:0.35785\n",
      "Train Epoch: 61 [320/1464 (22%)] ,Loss: 0.775607 ,Accuracy:84.41%, Train Mean IoU:0.36229\n",
      "Train Epoch: 61 [480/1464 (33%)] ,Loss: 0.964912 ,Accuracy:84.66%, Train Mean IoU:0.36334\n",
      "Train Epoch: 61 [640/1464 (44%)] ,Loss: 0.916842 ,Accuracy:84.96%, Train Mean IoU:0.36461\n",
      "Train Epoch: 61 [800/1464 (55%)] ,Loss: 0.658235 ,Accuracy:85.02%, Train Mean IoU:0.36569\n",
      "Train Epoch: 61 [960/1464 (66%)] ,Loss: 0.360275 ,Accuracy:85.29%, Train Mean IoU:0.36694\n",
      "Train Epoch: 61 [1120/1464 (77%)] ,Loss: 0.970542 ,Accuracy:85.09%, Train Mean IoU:0.36774\n",
      "Train Epoch: 61 [1280/1464 (87%)] ,Loss: 0.909924 ,Accuracy:84.98%, Train Mean IoU:0.36809\n",
      "Train Epoch: 61 [1440/1464 (98%)] ,Loss: 1.048271 ,Accuracy:84.88%, Train Mean IoU:0.36871\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 62 - - - - - - - -\n",
      "Train Epoch: 62 [160/1464 (11%)] ,Loss: 0.670476 ,Accuracy:83.71%, Train Mean IoU:0.36632\n",
      "Train Epoch: 62 [320/1464 (22%)] ,Loss: 0.688091 ,Accuracy:85.46%, Train Mean IoU:0.37148\n",
      "Train Epoch: 62 [480/1464 (33%)] ,Loss: 0.899217 ,Accuracy:85.33%, Train Mean IoU:0.37336\n",
      "Train Epoch: 62 [640/1464 (44%)] ,Loss: 0.691549 ,Accuracy:85.18%, Train Mean IoU:0.37232\n",
      "Train Epoch: 62 [800/1464 (55%)] ,Loss: 1.013929 ,Accuracy:84.80%, Train Mean IoU:0.36945\n",
      "Train Epoch: 62 [960/1464 (66%)] ,Loss: 0.925145 ,Accuracy:84.66%, Train Mean IoU:0.36644\n",
      "Train Epoch: 62 [1120/1464 (77%)] ,Loss: 0.715051 ,Accuracy:84.75%, Train Mean IoU:0.36786\n",
      "Train Epoch: 62 [1280/1464 (87%)] ,Loss: 1.062967 ,Accuracy:84.84%, Train Mean IoU:0.36809\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 62 [1440/1464 (98%)] ,Loss: 0.926895 ,Accuracy:84.92%, Train Mean IoU:0.36888\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 63 - - - - - - - -\n",
      "Train Epoch: 63 [160/1464 (11%)] ,Loss: 0.694627 ,Accuracy:86.57%, Train Mean IoU:0.37061\n",
      "Train Epoch: 63 [320/1464 (22%)] ,Loss: 0.869257 ,Accuracy:85.99%, Train Mean IoU:0.37407\n",
      "Train Epoch: 63 [480/1464 (33%)] ,Loss: 0.545929 ,Accuracy:85.79%, Train Mean IoU:0.37104\n",
      "Train Epoch: 63 [640/1464 (44%)] ,Loss: 0.678240 ,Accuracy:85.65%, Train Mean IoU:0.37179\n",
      "Train Epoch: 63 [800/1464 (55%)] ,Loss: 0.716967 ,Accuracy:85.82%, Train Mean IoU:0.36970\n",
      "Train Epoch: 63 [960/1464 (66%)] ,Loss: 0.716922 ,Accuracy:85.45%, Train Mean IoU:0.36931\n",
      "Train Epoch: 63 [1120/1464 (77%)] ,Loss: 0.821818 ,Accuracy:85.33%, Train Mean IoU:0.37091\n",
      "Train Epoch: 63 [1280/1464 (87%)] ,Loss: 0.706706 ,Accuracy:85.12%, Train Mean IoU:0.36993\n",
      "Train Epoch: 63 [1440/1464 (98%)] ,Loss: 0.933839 ,Accuracy:84.91%, Train Mean IoU:0.36901\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 64 - - - - - - - -\n",
      "Train Epoch: 64 [160/1464 (11%)] ,Loss: 1.020463 ,Accuracy:83.65%, Train Mean IoU:0.35921\n",
      "Train Epoch: 64 [320/1464 (22%)] ,Loss: 1.131401 ,Accuracy:84.79%, Train Mean IoU:0.37292\n",
      "Train Epoch: 64 [480/1464 (33%)] ,Loss: 0.866031 ,Accuracy:84.60%, Train Mean IoU:0.37049\n",
      "Train Epoch: 64 [640/1464 (44%)] ,Loss: 1.210316 ,Accuracy:84.64%, Train Mean IoU:0.36653\n",
      "Train Epoch: 64 [800/1464 (55%)] ,Loss: 0.708078 ,Accuracy:84.71%, Train Mean IoU:0.36780\n",
      "Train Epoch: 64 [960/1464 (66%)] ,Loss: 0.878157 ,Accuracy:84.71%, Train Mean IoU:0.36679\n",
      "Train Epoch: 64 [1120/1464 (77%)] ,Loss: 0.669376 ,Accuracy:84.75%, Train Mean IoU:0.36697\n",
      "Train Epoch: 64 [1280/1464 (87%)] ,Loss: 0.910284 ,Accuracy:84.80%, Train Mean IoU:0.36749\n",
      "Train Epoch: 64 [1440/1464 (98%)] ,Loss: 0.897575 ,Accuracy:84.89%, Train Mean IoU:0.36834\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 65 - - - - - - - -\n",
      "Train Epoch: 65 [160/1464 (11%)] ,Loss: 1.064486 ,Accuracy:85.39%, Train Mean IoU:0.37474\n",
      "Train Epoch: 65 [320/1464 (22%)] ,Loss: 0.791870 ,Accuracy:84.81%, Train Mean IoU:0.36489\n",
      "Train Epoch: 65 [480/1464 (33%)] ,Loss: 1.417964 ,Accuracy:84.16%, Train Mean IoU:0.36300\n",
      "Train Epoch: 65 [640/1464 (44%)] ,Loss: 0.975094 ,Accuracy:84.27%, Train Mean IoU:0.36510\n",
      "Train Epoch: 65 [800/1464 (55%)] ,Loss: 0.658583 ,Accuracy:84.66%, Train Mean IoU:0.36653\n",
      "Train Epoch: 65 [960/1464 (66%)] ,Loss: 0.673731 ,Accuracy:85.03%, Train Mean IoU:0.36816\n",
      "Train Epoch: 65 [1120/1464 (77%)] ,Loss: 1.263245 ,Accuracy:84.95%, Train Mean IoU:0.36883\n",
      "Train Epoch: 65 [1280/1464 (87%)] ,Loss: 0.635311 ,Accuracy:84.73%, Train Mean IoU:0.36685\n",
      "Train Epoch: 65 [1440/1464 (98%)] ,Loss: 0.674898 ,Accuracy:84.90%, Train Mean IoU:0.36856\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 66 - - - - - - - -\n",
      "Train Epoch: 66 [160/1464 (11%)] ,Loss: 0.882771 ,Accuracy:85.30%, Train Mean IoU:0.38068\n",
      "Train Epoch: 66 [320/1464 (22%)] ,Loss: 0.650755 ,Accuracy:85.02%, Train Mean IoU:0.37394\n",
      "Train Epoch: 66 [480/1464 (33%)] ,Loss: 0.519432 ,Accuracy:84.95%, Train Mean IoU:0.37265\n",
      "Train Epoch: 66 [640/1464 (44%)] ,Loss: 0.750732 ,Accuracy:84.86%, Train Mean IoU:0.36806\n",
      "Train Epoch: 66 [800/1464 (55%)] ,Loss: 0.478842 ,Accuracy:85.11%, Train Mean IoU:0.36834\n",
      "Train Epoch: 66 [960/1464 (66%)] ,Loss: 1.113791 ,Accuracy:85.13%, Train Mean IoU:0.36756\n",
      "Train Epoch: 66 [1120/1464 (77%)] ,Loss: 0.492388 ,Accuracy:85.00%, Train Mean IoU:0.36754\n",
      "Train Epoch: 66 [1280/1464 (87%)] ,Loss: 0.866441 ,Accuracy:84.93%, Train Mean IoU:0.36877\n",
      "Train Epoch: 66 [1440/1464 (98%)] ,Loss: 0.781321 ,Accuracy:84.90%, Train Mean IoU:0.36864\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 67 - - - - - - - -\n",
      "Train Epoch: 67 [160/1464 (11%)] ,Loss: 0.979557 ,Accuracy:86.19%, Train Mean IoU:0.36549\n",
      "Train Epoch: 67 [320/1464 (22%)] ,Loss: 0.809738 ,Accuracy:85.86%, Train Mean IoU:0.37166\n",
      "Train Epoch: 67 [480/1464 (33%)] ,Loss: 0.943509 ,Accuracy:85.30%, Train Mean IoU:0.37353\n",
      "Train Epoch: 67 [640/1464 (44%)] ,Loss: 0.621582 ,Accuracy:85.14%, Train Mean IoU:0.37097\n",
      "Train Epoch: 67 [800/1464 (55%)] ,Loss: 1.144432 ,Accuracy:85.04%, Train Mean IoU:0.37055\n",
      "Train Epoch: 67 [960/1464 (66%)] ,Loss: 0.819157 ,Accuracy:85.11%, Train Mean IoU:0.37102\n",
      "Train Epoch: 67 [1120/1464 (77%)] ,Loss: 0.871875 ,Accuracy:85.13%, Train Mean IoU:0.36941\n",
      "Train Epoch: 67 [1280/1464 (87%)] ,Loss: 0.910104 ,Accuracy:85.27%, Train Mean IoU:0.36985\n",
      "Train Epoch: 67 [1440/1464 (98%)] ,Loss: 1.085589 ,Accuracy:84.89%, Train Mean IoU:0.36843\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 68 - - - - - - - -\n",
      "Train Epoch: 68 [160/1464 (11%)] ,Loss: 1.294625 ,Accuracy:82.22%, Train Mean IoU:0.36131\n",
      "Train Epoch: 68 [320/1464 (22%)] ,Loss: 0.875166 ,Accuracy:83.86%, Train Mean IoU:0.36572\n",
      "Train Epoch: 68 [480/1464 (33%)] ,Loss: 0.817101 ,Accuracy:84.22%, Train Mean IoU:0.36654\n",
      "Train Epoch: 68 [640/1464 (44%)] ,Loss: 0.567501 ,Accuracy:84.48%, Train Mean IoU:0.36624\n",
      "Train Epoch: 68 [800/1464 (55%)] ,Loss: 0.976684 ,Accuracy:84.10%, Train Mean IoU:0.36408\n",
      "Train Epoch: 68 [960/1464 (66%)] ,Loss: 0.705932 ,Accuracy:84.44%, Train Mean IoU:0.36569\n",
      "Train Epoch: 68 [1120/1464 (77%)] ,Loss: 0.961421 ,Accuracy:84.42%, Train Mean IoU:0.36675\n",
      "Train Epoch: 68 [1280/1464 (87%)] ,Loss: 1.116198 ,Accuracy:84.63%, Train Mean IoU:0.36707\n",
      "Train Epoch: 68 [1440/1464 (98%)] ,Loss: 1.020878 ,Accuracy:84.85%, Train Mean IoU:0.36822\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 69 - - - - - - - -\n",
      "Train Epoch: 69 [160/1464 (11%)] ,Loss: 0.804678 ,Accuracy:84.20%, Train Mean IoU:0.36778\n",
      "Train Epoch: 69 [320/1464 (22%)] ,Loss: 0.917650 ,Accuracy:83.54%, Train Mean IoU:0.36556\n",
      "Train Epoch: 69 [480/1464 (33%)] ,Loss: 0.503100 ,Accuracy:83.92%, Train Mean IoU:0.36989\n",
      "Train Epoch: 69 [640/1464 (44%)] ,Loss: 0.809832 ,Accuracy:83.98%, Train Mean IoU:0.36555\n",
      "Train Epoch: 69 [800/1464 (55%)] ,Loss: 0.525082 ,Accuracy:84.39%, Train Mean IoU:0.36654\n",
      "Train Epoch: 69 [960/1464 (66%)] ,Loss: 0.922643 ,Accuracy:84.48%, Train Mean IoU:0.36804\n",
      "Train Epoch: 69 [1120/1464 (77%)] ,Loss: 1.085333 ,Accuracy:84.73%, Train Mean IoU:0.36880\n",
      "Train Epoch: 69 [1280/1464 (87%)] ,Loss: 0.724303 ,Accuracy:84.70%, Train Mean IoU:0.36901\n",
      "Train Epoch: 69 [1440/1464 (98%)] ,Loss: 0.629689 ,Accuracy:84.88%, Train Mean IoU:0.36886\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 70 - - - - - - - -\n",
      "Train Epoch: 70 [160/1464 (11%)] ,Loss: 0.914532 ,Accuracy:85.54%, Train Mean IoU:0.38056\n",
      "Train Epoch: 70 [320/1464 (22%)] ,Loss: 0.854795 ,Accuracy:85.16%, Train Mean IoU:0.37281\n",
      "Train Epoch: 70 [480/1464 (33%)] ,Loss: 0.748501 ,Accuracy:85.47%, Train Mean IoU:0.37185\n",
      "Train Epoch: 70 [640/1464 (44%)] ,Loss: 0.848021 ,Accuracy:85.24%, Train Mean IoU:0.37029\n",
      "Train Epoch: 70 [800/1464 (55%)] ,Loss: 0.576094 ,Accuracy:85.28%, Train Mean IoU:0.37186\n",
      "Train Epoch: 70 [960/1464 (66%)] ,Loss: 0.719173 ,Accuracy:85.08%, Train Mean IoU:0.36846\n",
      "Train Epoch: 70 [1120/1464 (77%)] ,Loss: 0.855488 ,Accuracy:84.87%, Train Mean IoU:0.36780\n",
      "Train Epoch: 70 [1280/1464 (87%)] ,Loss: 0.579842 ,Accuracy:84.99%, Train Mean IoU:0.36804\n",
      "Train Epoch: 70 [1440/1464 (98%)] ,Loss: 0.566312 ,Accuracy:84.86%, Train Mean IoU:0.36844\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 71 - - - - - - - -\n",
      "Train Epoch: 71 [160/1464 (11%)] ,Loss: 1.178677 ,Accuracy:83.57%, Train Mean IoU:0.35119\n",
      "Train Epoch: 71 [320/1464 (22%)] ,Loss: 0.805232 ,Accuracy:84.37%, Train Mean IoU:0.36315\n",
      "Train Epoch: 71 [480/1464 (33%)] ,Loss: 0.872376 ,Accuracy:84.59%, Train Mean IoU:0.36736\n",
      "Train Epoch: 71 [640/1464 (44%)] ,Loss: 0.675470 ,Accuracy:84.49%, Train Mean IoU:0.36676\n",
      "Train Epoch: 71 [800/1464 (55%)] ,Loss: 0.604905 ,Accuracy:84.82%, Train Mean IoU:0.36728\n",
      "Train Epoch: 71 [960/1464 (66%)] ,Loss: 0.950505 ,Accuracy:84.92%, Train Mean IoU:0.36792\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 71 [1120/1464 (77%)] ,Loss: 0.917131 ,Accuracy:85.09%, Train Mean IoU:0.36987\n",
      "Train Epoch: 71 [1280/1464 (87%)] ,Loss: 0.886291 ,Accuracy:85.01%, Train Mean IoU:0.36922\n",
      "Train Epoch: 71 [1440/1464 (98%)] ,Loss: 0.717558 ,Accuracy:84.86%, Train Mean IoU:0.36878\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 72 - - - - - - - -\n",
      "Train Epoch: 72 [160/1464 (11%)] ,Loss: 0.634068 ,Accuracy:85.42%, Train Mean IoU:0.38143\n",
      "Train Epoch: 72 [320/1464 (22%)] ,Loss: 0.867378 ,Accuracy:85.28%, Train Mean IoU:0.37265\n",
      "Train Epoch: 72 [480/1464 (33%)] ,Loss: 0.984190 ,Accuracy:85.69%, Train Mean IoU:0.37546\n",
      "Train Epoch: 72 [640/1464 (44%)] ,Loss: 0.676903 ,Accuracy:85.71%, Train Mean IoU:0.37421\n",
      "Train Epoch: 72 [800/1464 (55%)] ,Loss: 0.820944 ,Accuracy:85.62%, Train Mean IoU:0.37103\n",
      "Train Epoch: 72 [960/1464 (66%)] ,Loss: 0.677773 ,Accuracy:85.49%, Train Mean IoU:0.37048\n",
      "Train Epoch: 72 [1120/1464 (77%)] ,Loss: 0.900896 ,Accuracy:85.21%, Train Mean IoU:0.36924\n",
      "Train Epoch: 72 [1280/1464 (87%)] ,Loss: 0.975487 ,Accuracy:85.16%, Train Mean IoU:0.37078\n",
      "Train Epoch: 72 [1440/1464 (98%)] ,Loss: 0.976019 ,Accuracy:84.86%, Train Mean IoU:0.36894\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 73 - - - - - - - -\n",
      "Train Epoch: 73 [160/1464 (11%)] ,Loss: 0.776569 ,Accuracy:85.02%, Train Mean IoU:0.36673\n",
      "Train Epoch: 73 [320/1464 (22%)] ,Loss: 0.602157 ,Accuracy:84.65%, Train Mean IoU:0.36393\n",
      "Train Epoch: 73 [480/1464 (33%)] ,Loss: 0.897799 ,Accuracy:85.24%, Train Mean IoU:0.36747\n",
      "Train Epoch: 73 [640/1464 (44%)] ,Loss: 0.768218 ,Accuracy:85.48%, Train Mean IoU:0.36815\n",
      "Train Epoch: 73 [800/1464 (55%)] ,Loss: 0.892739 ,Accuracy:85.00%, Train Mean IoU:0.36806\n",
      "Train Epoch: 73 [960/1464 (66%)] ,Loss: 0.725295 ,Accuracy:85.08%, Train Mean IoU:0.36997\n",
      "Train Epoch: 73 [1120/1464 (77%)] ,Loss: 0.639403 ,Accuracy:85.05%, Train Mean IoU:0.37051\n",
      "Train Epoch: 73 [1280/1464 (87%)] ,Loss: 0.996805 ,Accuracy:84.94%, Train Mean IoU:0.36919\n",
      "Train Epoch: 73 [1440/1464 (98%)] ,Loss: 0.847222 ,Accuracy:84.90%, Train Mean IoU:0.36867\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 74 - - - - - - - -\n",
      "Train Epoch: 74 [160/1464 (11%)] ,Loss: 1.034315 ,Accuracy:82.76%, Train Mean IoU:0.36313\n",
      "Train Epoch: 74 [320/1464 (22%)] ,Loss: 0.861173 ,Accuracy:84.19%, Train Mean IoU:0.36682\n",
      "Train Epoch: 74 [480/1464 (33%)] ,Loss: 0.896977 ,Accuracy:84.73%, Train Mean IoU:0.36881\n",
      "Train Epoch: 74 [640/1464 (44%)] ,Loss: 1.286868 ,Accuracy:84.77%, Train Mean IoU:0.36758\n",
      "Train Epoch: 74 [800/1464 (55%)] ,Loss: 0.590959 ,Accuracy:84.84%, Train Mean IoU:0.36764\n",
      "Train Epoch: 74 [960/1464 (66%)] ,Loss: 0.735799 ,Accuracy:84.87%, Train Mean IoU:0.36886\n",
      "Train Epoch: 74 [1120/1464 (77%)] ,Loss: 0.559421 ,Accuracy:84.91%, Train Mean IoU:0.36963\n",
      "Train Epoch: 74 [1280/1464 (87%)] ,Loss: 0.763798 ,Accuracy:84.75%, Train Mean IoU:0.36795\n",
      "Train Epoch: 74 [1440/1464 (98%)] ,Loss: 0.815783 ,Accuracy:84.86%, Train Mean IoU:0.36893\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 75 - - - - - - - -\n",
      "Train Epoch: 75 [160/1464 (11%)] ,Loss: 0.717387 ,Accuracy:84.91%, Train Mean IoU:0.36807\n",
      "Train Epoch: 75 [320/1464 (22%)] ,Loss: 0.806789 ,Accuracy:85.01%, Train Mean IoU:0.36638\n",
      "Train Epoch: 75 [480/1464 (33%)] ,Loss: 0.613982 ,Accuracy:85.50%, Train Mean IoU:0.37186\n",
      "Train Epoch: 75 [640/1464 (44%)] ,Loss: 0.642606 ,Accuracy:85.43%, Train Mean IoU:0.37113\n",
      "Train Epoch: 75 [800/1464 (55%)] ,Loss: 0.752663 ,Accuracy:85.39%, Train Mean IoU:0.37280\n",
      "Train Epoch: 75 [960/1464 (66%)] ,Loss: 0.597119 ,Accuracy:85.27%, Train Mean IoU:0.37188\n",
      "Train Epoch: 75 [1120/1464 (77%)] ,Loss: 0.482641 ,Accuracy:85.17%, Train Mean IoU:0.37094\n",
      "Train Epoch: 75 [1280/1464 (87%)] ,Loss: 1.078427 ,Accuracy:85.04%, Train Mean IoU:0.36932\n",
      "Train Epoch: 75 [1440/1464 (98%)] ,Loss: 1.019245 ,Accuracy:84.92%, Train Mean IoU:0.36844\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 76 - - - - - - - -\n",
      "Train Epoch: 76 [160/1464 (11%)] ,Loss: 0.902797 ,Accuracy:84.59%, Train Mean IoU:0.36409\n",
      "Train Epoch: 76 [320/1464 (22%)] ,Loss: 1.198580 ,Accuracy:85.02%, Train Mean IoU:0.37015\n",
      "Train Epoch: 76 [480/1464 (33%)] ,Loss: 0.957497 ,Accuracy:84.90%, Train Mean IoU:0.36923\n",
      "Train Epoch: 76 [640/1464 (44%)] ,Loss: 1.377699 ,Accuracy:84.80%, Train Mean IoU:0.36767\n",
      "Train Epoch: 76 [800/1464 (55%)] ,Loss: 0.628664 ,Accuracy:84.41%, Train Mean IoU:0.36805\n",
      "Train Epoch: 76 [960/1464 (66%)] ,Loss: 0.710288 ,Accuracy:84.73%, Train Mean IoU:0.36791\n",
      "Train Epoch: 76 [1120/1464 (77%)] ,Loss: 0.462502 ,Accuracy:84.81%, Train Mean IoU:0.36907\n",
      "Train Epoch: 76 [1280/1464 (87%)] ,Loss: 0.613560 ,Accuracy:84.99%, Train Mean IoU:0.37049\n",
      "Train Epoch: 76 [1440/1464 (98%)] ,Loss: 0.961474 ,Accuracy:84.85%, Train Mean IoU:0.36864\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 77 - - - - - - - -\n",
      "Train Epoch: 77 [160/1464 (11%)] ,Loss: 0.877557 ,Accuracy:82.34%, Train Mean IoU:0.35977\n",
      "Train Epoch: 77 [320/1464 (22%)] ,Loss: 0.850793 ,Accuracy:84.03%, Train Mean IoU:0.36281\n",
      "Train Epoch: 77 [480/1464 (33%)] ,Loss: 0.928744 ,Accuracy:83.76%, Train Mean IoU:0.36417\n",
      "Train Epoch: 77 [640/1464 (44%)] ,Loss: 0.857925 ,Accuracy:84.63%, Train Mean IoU:0.36857\n",
      "Train Epoch: 77 [800/1464 (55%)] ,Loss: 0.882914 ,Accuracy:84.47%, Train Mean IoU:0.36835\n",
      "Train Epoch: 77 [960/1464 (66%)] ,Loss: 0.534534 ,Accuracy:84.68%, Train Mean IoU:0.36911\n",
      "Train Epoch: 77 [1120/1464 (77%)] ,Loss: 0.530649 ,Accuracy:84.85%, Train Mean IoU:0.36891\n",
      "Train Epoch: 77 [1280/1464 (87%)] ,Loss: 0.939411 ,Accuracy:84.99%, Train Mean IoU:0.36942\n",
      "Train Epoch: 77 [1440/1464 (98%)] ,Loss: 0.732832 ,Accuracy:84.96%, Train Mean IoU:0.36912\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 78 - - - - - - - -\n",
      "Train Epoch: 78 [160/1464 (11%)] ,Loss: 0.981760 ,Accuracy:84.61%, Train Mean IoU:0.37151\n",
      "Train Epoch: 78 [320/1464 (22%)] ,Loss: 0.960904 ,Accuracy:84.35%, Train Mean IoU:0.36874\n",
      "Train Epoch: 78 [480/1464 (33%)] ,Loss: 0.855837 ,Accuracy:84.53%, Train Mean IoU:0.37281\n",
      "Train Epoch: 78 [640/1464 (44%)] ,Loss: 0.991543 ,Accuracy:84.65%, Train Mean IoU:0.37254\n",
      "Train Epoch: 78 [800/1464 (55%)] ,Loss: 0.669415 ,Accuracy:84.64%, Train Mean IoU:0.37294\n",
      "Train Epoch: 78 [960/1464 (66%)] ,Loss: 0.662694 ,Accuracy:84.63%, Train Mean IoU:0.37069\n",
      "Train Epoch: 78 [1120/1464 (77%)] ,Loss: 0.471596 ,Accuracy:84.77%, Train Mean IoU:0.37195\n",
      "Train Epoch: 78 [1280/1464 (87%)] ,Loss: 0.763615 ,Accuracy:84.91%, Train Mean IoU:0.37041\n",
      "Train Epoch: 78 [1440/1464 (98%)] ,Loss: 0.798110 ,Accuracy:84.93%, Train Mean IoU:0.36905\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 79 - - - - - - - -\n",
      "Train Epoch: 79 [160/1464 (11%)] ,Loss: 0.711743 ,Accuracy:84.93%, Train Mean IoU:0.36083\n",
      "Train Epoch: 79 [320/1464 (22%)] ,Loss: 1.245556 ,Accuracy:84.32%, Train Mean IoU:0.35913\n",
      "Train Epoch: 79 [480/1464 (33%)] ,Loss: 0.751470 ,Accuracy:84.59%, Train Mean IoU:0.36088\n",
      "Train Epoch: 79 [640/1464 (44%)] ,Loss: 0.647791 ,Accuracy:84.57%, Train Mean IoU:0.36463\n",
      "Train Epoch: 79 [800/1464 (55%)] ,Loss: 0.755323 ,Accuracy:84.56%, Train Mean IoU:0.36515\n",
      "Train Epoch: 79 [960/1464 (66%)] ,Loss: 0.852037 ,Accuracy:84.78%, Train Mean IoU:0.36735\n",
      "Train Epoch: 79 [1120/1464 (77%)] ,Loss: 0.756190 ,Accuracy:84.81%, Train Mean IoU:0.36793\n",
      "Train Epoch: 79 [1280/1464 (87%)] ,Loss: 0.510379 ,Accuracy:84.92%, Train Mean IoU:0.36917\n",
      "Train Epoch: 79 [1440/1464 (98%)] ,Loss: 0.604915 ,Accuracy:84.90%, Train Mean IoU:0.36890\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 80 - - - - - - - -\n",
      "Train Epoch: 80 [160/1464 (11%)] ,Loss: 1.016683 ,Accuracy:85.65%, Train Mean IoU:0.37816\n",
      "Train Epoch: 80 [320/1464 (22%)] ,Loss: 0.597940 ,Accuracy:84.66%, Train Mean IoU:0.37125\n",
      "Train Epoch: 80 [480/1464 (33%)] ,Loss: 0.799761 ,Accuracy:84.81%, Train Mean IoU:0.36739\n",
      "Train Epoch: 80 [640/1464 (44%)] ,Loss: 0.612408 ,Accuracy:84.65%, Train Mean IoU:0.36406\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 80 [800/1464 (55%)] ,Loss: 0.793750 ,Accuracy:85.16%, Train Mean IoU:0.36849\n",
      "Train Epoch: 80 [960/1464 (66%)] ,Loss: 0.749034 ,Accuracy:85.06%, Train Mean IoU:0.36867\n",
      "Train Epoch: 80 [1120/1464 (77%)] ,Loss: 0.679658 ,Accuracy:85.18%, Train Mean IoU:0.37082\n",
      "Train Epoch: 80 [1280/1464 (87%)] ,Loss: 0.657960 ,Accuracy:85.11%, Train Mean IoU:0.37059\n",
      "Train Epoch: 80 [1440/1464 (98%)] ,Loss: 1.241357 ,Accuracy:84.89%, Train Mean IoU:0.36867\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 81 - - - - - - - -\n",
      "Train Epoch: 81 [160/1464 (11%)] ,Loss: 0.732551 ,Accuracy:84.60%, Train Mean IoU:0.36179\n",
      "Train Epoch: 81 [320/1464 (22%)] ,Loss: 0.571267 ,Accuracy:84.95%, Train Mean IoU:0.36495\n",
      "Train Epoch: 81 [480/1464 (33%)] ,Loss: 0.777130 ,Accuracy:84.95%, Train Mean IoU:0.36855\n",
      "Train Epoch: 81 [640/1464 (44%)] ,Loss: 0.815974 ,Accuracy:84.94%, Train Mean IoU:0.36800\n",
      "Train Epoch: 81 [800/1464 (55%)] ,Loss: 1.095025 ,Accuracy:84.69%, Train Mean IoU:0.36853\n",
      "Train Epoch: 81 [960/1464 (66%)] ,Loss: 0.577929 ,Accuracy:84.87%, Train Mean IoU:0.36835\n",
      "Train Epoch: 81 [1120/1464 (77%)] ,Loss: 0.553873 ,Accuracy:84.80%, Train Mean IoU:0.36814\n",
      "Train Epoch: 81 [1280/1464 (87%)] ,Loss: 0.687727 ,Accuracy:84.86%, Train Mean IoU:0.36789\n",
      "Train Epoch: 81 [1440/1464 (98%)] ,Loss: 0.788920 ,Accuracy:84.90%, Train Mean IoU:0.36908\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 82 - - - - - - - -\n",
      "Train Epoch: 82 [160/1464 (11%)] ,Loss: 0.615362 ,Accuracy:84.40%, Train Mean IoU:0.37014\n",
      "Train Epoch: 82 [320/1464 (22%)] ,Loss: 0.907588 ,Accuracy:84.00%, Train Mean IoU:0.36255\n",
      "Train Epoch: 82 [480/1464 (33%)] ,Loss: 1.222671 ,Accuracy:84.49%, Train Mean IoU:0.36678\n",
      "Train Epoch: 82 [640/1464 (44%)] ,Loss: 0.773376 ,Accuracy:84.55%, Train Mean IoU:0.36586\n",
      "Train Epoch: 82 [800/1464 (55%)] ,Loss: 0.979221 ,Accuracy:84.69%, Train Mean IoU:0.36653\n",
      "Train Epoch: 82 [960/1464 (66%)] ,Loss: 0.641590 ,Accuracy:84.77%, Train Mean IoU:0.36829\n",
      "Train Epoch: 82 [1120/1464 (77%)] ,Loss: 0.766441 ,Accuracy:84.85%, Train Mean IoU:0.36969\n",
      "Train Epoch: 82 [1280/1464 (87%)] ,Loss: 0.589465 ,Accuracy:84.97%, Train Mean IoU:0.37092\n",
      "Train Epoch: 82 [1440/1464 (98%)] ,Loss: 0.685728 ,Accuracy:84.87%, Train Mean IoU:0.36874\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 83 - - - - - - - -\n",
      "Train Epoch: 83 [160/1464 (11%)] ,Loss: 0.614872 ,Accuracy:84.54%, Train Mean IoU:0.37467\n",
      "Train Epoch: 83 [320/1464 (22%)] ,Loss: 0.675212 ,Accuracy:84.70%, Train Mean IoU:0.37727\n",
      "Train Epoch: 83 [480/1464 (33%)] ,Loss: 0.791677 ,Accuracy:84.92%, Train Mean IoU:0.37582\n",
      "Train Epoch: 83 [640/1464 (44%)] ,Loss: 0.850738 ,Accuracy:84.78%, Train Mean IoU:0.37386\n",
      "Train Epoch: 83 [800/1464 (55%)] ,Loss: 0.767804 ,Accuracy:85.11%, Train Mean IoU:0.37354\n",
      "Train Epoch: 83 [960/1464 (66%)] ,Loss: 0.528178 ,Accuracy:85.04%, Train Mean IoU:0.37009\n",
      "Train Epoch: 83 [1120/1464 (77%)] ,Loss: 0.491792 ,Accuracy:85.09%, Train Mean IoU:0.37040\n",
      "Train Epoch: 83 [1280/1464 (87%)] ,Loss: 1.013180 ,Accuracy:84.92%, Train Mean IoU:0.36900\n",
      "Train Epoch: 83 [1440/1464 (98%)] ,Loss: 1.406446 ,Accuracy:84.92%, Train Mean IoU:0.36828\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 84 - - - - - - - -\n",
      "Train Epoch: 84 [160/1464 (11%)] ,Loss: 0.761531 ,Accuracy:86.40%, Train Mean IoU:0.38226\n",
      "Train Epoch: 84 [320/1464 (22%)] ,Loss: 0.681121 ,Accuracy:85.29%, Train Mean IoU:0.37576\n",
      "Train Epoch: 84 [480/1464 (33%)] ,Loss: 1.272531 ,Accuracy:85.08%, Train Mean IoU:0.37278\n",
      "Train Epoch: 84 [640/1464 (44%)] ,Loss: 0.811997 ,Accuracy:85.03%, Train Mean IoU:0.36905\n",
      "Train Epoch: 84 [800/1464 (55%)] ,Loss: 0.558636 ,Accuracy:85.17%, Train Mean IoU:0.36864\n",
      "Train Epoch: 84 [960/1464 (66%)] ,Loss: 0.627108 ,Accuracy:84.96%, Train Mean IoU:0.36778\n",
      "Train Epoch: 84 [1120/1464 (77%)] ,Loss: 0.643218 ,Accuracy:85.13%, Train Mean IoU:0.36857\n",
      "Train Epoch: 84 [1280/1464 (87%)] ,Loss: 0.809971 ,Accuracy:85.13%, Train Mean IoU:0.36989\n",
      "Train Epoch: 84 [1440/1464 (98%)] ,Loss: 0.883508 ,Accuracy:84.86%, Train Mean IoU:0.36860\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 85 - - - - - - - -\n",
      "Train Epoch: 85 [160/1464 (11%)] ,Loss: 0.777258 ,Accuracy:84.52%, Train Mean IoU:0.36164\n",
      "Train Epoch: 85 [320/1464 (22%)] ,Loss: 0.612053 ,Accuracy:84.81%, Train Mean IoU:0.36471\n",
      "Train Epoch: 85 [480/1464 (33%)] ,Loss: 0.981182 ,Accuracy:84.63%, Train Mean IoU:0.36091\n",
      "Train Epoch: 85 [640/1464 (44%)] ,Loss: 0.707975 ,Accuracy:84.63%, Train Mean IoU:0.36406\n",
      "Train Epoch: 85 [800/1464 (55%)] ,Loss: 1.053005 ,Accuracy:85.01%, Train Mean IoU:0.36953\n",
      "Train Epoch: 85 [960/1464 (66%)] ,Loss: 0.894414 ,Accuracy:84.86%, Train Mean IoU:0.36835\n",
      "Train Epoch: 85 [1120/1464 (77%)] ,Loss: 0.514856 ,Accuracy:85.01%, Train Mean IoU:0.36929\n",
      "Train Epoch: 85 [1280/1464 (87%)] ,Loss: 0.579596 ,Accuracy:85.15%, Train Mean IoU:0.37030\n",
      "Train Epoch: 85 [1440/1464 (98%)] ,Loss: 0.884908 ,Accuracy:84.95%, Train Mean IoU:0.36876\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 86 - - - - - - - -\n",
      "Train Epoch: 86 [160/1464 (11%)] ,Loss: 0.607465 ,Accuracy:84.41%, Train Mean IoU:0.36002\n",
      "Train Epoch: 86 [320/1464 (22%)] ,Loss: 0.517428 ,Accuracy:84.46%, Train Mean IoU:0.36586\n",
      "Train Epoch: 86 [480/1464 (33%)] ,Loss: 0.692661 ,Accuracy:84.36%, Train Mean IoU:0.36699\n",
      "Train Epoch: 86 [640/1464 (44%)] ,Loss: 0.897732 ,Accuracy:84.62%, Train Mean IoU:0.36754\n",
      "Train Epoch: 86 [800/1464 (55%)] ,Loss: 1.200695 ,Accuracy:84.58%, Train Mean IoU:0.36596\n",
      "Train Epoch: 86 [960/1464 (66%)] ,Loss: 1.301636 ,Accuracy:84.95%, Train Mean IoU:0.36727\n",
      "Train Epoch: 86 [1120/1464 (77%)] ,Loss: 1.068132 ,Accuracy:84.91%, Train Mean IoU:0.36888\n",
      "Train Epoch: 86 [1280/1464 (87%)] ,Loss: 0.426331 ,Accuracy:85.01%, Train Mean IoU:0.36861\n",
      "Train Epoch: 86 [1440/1464 (98%)] ,Loss: 1.136130 ,Accuracy:84.94%, Train Mean IoU:0.36865\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 87 - - - - - - - -\n",
      "Train Epoch: 87 [160/1464 (11%)] ,Loss: 0.960169 ,Accuracy:84.31%, Train Mean IoU:0.36539\n",
      "Train Epoch: 87 [320/1464 (22%)] ,Loss: 0.666659 ,Accuracy:84.85%, Train Mean IoU:0.37532\n",
      "Train Epoch: 87 [480/1464 (33%)] ,Loss: 0.883946 ,Accuracy:85.16%, Train Mean IoU:0.37653\n",
      "Train Epoch: 87 [640/1464 (44%)] ,Loss: 0.669595 ,Accuracy:85.21%, Train Mean IoU:0.37497\n",
      "Train Epoch: 87 [800/1464 (55%)] ,Loss: 0.870420 ,Accuracy:85.25%, Train Mean IoU:0.37601\n",
      "Train Epoch: 87 [960/1464 (66%)] ,Loss: 0.835604 ,Accuracy:84.81%, Train Mean IoU:0.37125\n",
      "Train Epoch: 87 [1120/1464 (77%)] ,Loss: 0.897485 ,Accuracy:84.92%, Train Mean IoU:0.37095\n",
      "Train Epoch: 87 [1280/1464 (87%)] ,Loss: 0.979958 ,Accuracy:84.88%, Train Mean IoU:0.36848\n",
      "Train Epoch: 87 [1440/1464 (98%)] ,Loss: 0.626617 ,Accuracy:84.89%, Train Mean IoU:0.36877\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 88 - - - - - - - -\n",
      "Train Epoch: 88 [160/1464 (11%)] ,Loss: 0.613947 ,Accuracy:86.08%, Train Mean IoU:0.37773\n",
      "Train Epoch: 88 [320/1464 (22%)] ,Loss: 0.740414 ,Accuracy:85.79%, Train Mean IoU:0.38044\n",
      "Train Epoch: 88 [480/1464 (33%)] ,Loss: 0.681153 ,Accuracy:85.69%, Train Mean IoU:0.37809\n",
      "Train Epoch: 88 [640/1464 (44%)] ,Loss: 1.120316 ,Accuracy:85.34%, Train Mean IoU:0.37408\n",
      "Train Epoch: 88 [800/1464 (55%)] ,Loss: 0.647961 ,Accuracy:85.16%, Train Mean IoU:0.36899\n",
      "Train Epoch: 88 [960/1464 (66%)] ,Loss: 0.622764 ,Accuracy:84.99%, Train Mean IoU:0.36857\n",
      "Train Epoch: 88 [1120/1464 (77%)] ,Loss: 0.983828 ,Accuracy:85.11%, Train Mean IoU:0.36963\n",
      "Train Epoch: 88 [1280/1464 (87%)] ,Loss: 0.927404 ,Accuracy:84.99%, Train Mean IoU:0.36933\n",
      "Train Epoch: 88 [1440/1464 (98%)] ,Loss: 0.838183 ,Accuracy:84.97%, Train Mean IoU:0.36937\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 89 - - - - - - - -\n",
      "Train Epoch: 89 [160/1464 (11%)] ,Loss: 0.848188 ,Accuracy:82.91%, Train Mean IoU:0.36131\n",
      "Train Epoch: 89 [320/1464 (22%)] ,Loss: 0.779231 ,Accuracy:84.82%, Train Mean IoU:0.36620\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 89 [480/1464 (33%)] ,Loss: 0.659723 ,Accuracy:84.34%, Train Mean IoU:0.36157\n",
      "Train Epoch: 89 [640/1464 (44%)] ,Loss: 0.622848 ,Accuracy:84.42%, Train Mean IoU:0.36515\n",
      "Train Epoch: 89 [800/1464 (55%)] ,Loss: 0.901543 ,Accuracy:84.65%, Train Mean IoU:0.36674\n",
      "Train Epoch: 89 [960/1464 (66%)] ,Loss: 1.276019 ,Accuracy:84.77%, Train Mean IoU:0.36703\n",
      "Train Epoch: 89 [1120/1464 (77%)] ,Loss: 0.803418 ,Accuracy:84.72%, Train Mean IoU:0.36836\n",
      "Train Epoch: 89 [1280/1464 (87%)] ,Loss: 0.547696 ,Accuracy:84.73%, Train Mean IoU:0.36663\n",
      "Train Epoch: 89 [1440/1464 (98%)] ,Loss: 0.670748 ,Accuracy:84.81%, Train Mean IoU:0.36810\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 90 - - - - - - - -\n",
      "Train Epoch: 90 [160/1464 (11%)] ,Loss: 0.575398 ,Accuracy:83.96%, Train Mean IoU:0.36031\n",
      "Train Epoch: 90 [320/1464 (22%)] ,Loss: 1.012705 ,Accuracy:84.35%, Train Mean IoU:0.36231\n",
      "Train Epoch: 90 [480/1464 (33%)] ,Loss: 0.511496 ,Accuracy:84.28%, Train Mean IoU:0.36421\n",
      "Train Epoch: 90 [640/1464 (44%)] ,Loss: 0.694332 ,Accuracy:84.31%, Train Mean IoU:0.36110\n",
      "Train Epoch: 90 [800/1464 (55%)] ,Loss: 0.602020 ,Accuracy:84.58%, Train Mean IoU:0.36498\n",
      "Train Epoch: 90 [960/1464 (66%)] ,Loss: 0.701443 ,Accuracy:84.84%, Train Mean IoU:0.36544\n",
      "Train Epoch: 90 [1120/1464 (77%)] ,Loss: 0.478490 ,Accuracy:84.81%, Train Mean IoU:0.36552\n",
      "Train Epoch: 90 [1280/1464 (87%)] ,Loss: 0.597014 ,Accuracy:84.94%, Train Mean IoU:0.36676\n",
      "Train Epoch: 90 [1440/1464 (98%)] ,Loss: 0.995967 ,Accuracy:84.91%, Train Mean IoU:0.36861\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 91 - - - - - - - -\n",
      "LR is set to 0.000125\n",
      "Train Epoch: 91 [160/1464 (11%)] ,Loss: 0.684645 ,Accuracy:86.23%, Train Mean IoU:0.37609\n",
      "Train Epoch: 91 [320/1464 (22%)] ,Loss: 0.894475 ,Accuracy:85.25%, Train Mean IoU:0.37084\n",
      "Train Epoch: 91 [480/1464 (33%)] ,Loss: 0.968466 ,Accuracy:85.12%, Train Mean IoU:0.36749\n",
      "Train Epoch: 91 [640/1464 (44%)] ,Loss: 1.005396 ,Accuracy:85.60%, Train Mean IoU:0.36998\n",
      "Train Epoch: 91 [800/1464 (55%)] ,Loss: 0.735911 ,Accuracy:85.30%, Train Mean IoU:0.36957\n",
      "Train Epoch: 91 [960/1464 (66%)] ,Loss: 0.840413 ,Accuracy:85.13%, Train Mean IoU:0.36859\n",
      "Train Epoch: 91 [1120/1464 (77%)] ,Loss: 1.040132 ,Accuracy:84.88%, Train Mean IoU:0.36819\n",
      "Train Epoch: 91 [1280/1464 (87%)] ,Loss: 0.799666 ,Accuracy:85.03%, Train Mean IoU:0.36876\n",
      "Train Epoch: 91 [1440/1464 (98%)] ,Loss: 0.832910 ,Accuracy:84.91%, Train Mean IoU:0.36859\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 92 - - - - - - - -\n",
      "Train Epoch: 92 [160/1464 (11%)] ,Loss: 1.301040 ,Accuracy:83.74%, Train Mean IoU:0.35720\n",
      "Train Epoch: 92 [320/1464 (22%)] ,Loss: 0.897098 ,Accuracy:83.58%, Train Mean IoU:0.35924\n",
      "Train Epoch: 92 [480/1464 (33%)] ,Loss: 0.867232 ,Accuracy:84.03%, Train Mean IoU:0.36521\n",
      "Train Epoch: 92 [640/1464 (44%)] ,Loss: 0.749918 ,Accuracy:84.15%, Train Mean IoU:0.36708\n",
      "Train Epoch: 92 [800/1464 (55%)] ,Loss: 1.035264 ,Accuracy:84.55%, Train Mean IoU:0.36760\n",
      "Train Epoch: 92 [960/1464 (66%)] ,Loss: 0.676588 ,Accuracy:84.37%, Train Mean IoU:0.36689\n",
      "Train Epoch: 92 [1120/1464 (77%)] ,Loss: 0.901455 ,Accuracy:84.69%, Train Mean IoU:0.36659\n",
      "Train Epoch: 92 [1280/1464 (87%)] ,Loss: 0.967365 ,Accuracy:84.78%, Train Mean IoU:0.36754\n",
      "Train Epoch: 92 [1440/1464 (98%)] ,Loss: 0.574420 ,Accuracy:84.91%, Train Mean IoU:0.36877\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 93 - - - - - - - -\n",
      "Train Epoch: 93 [160/1464 (11%)] ,Loss: 0.815021 ,Accuracy:85.37%, Train Mean IoU:0.37024\n",
      "Train Epoch: 93 [320/1464 (22%)] ,Loss: 1.051602 ,Accuracy:85.29%, Train Mean IoU:0.37609\n",
      "Train Epoch: 93 [480/1464 (33%)] ,Loss: 1.163961 ,Accuracy:85.40%, Train Mean IoU:0.37311\n",
      "Train Epoch: 93 [640/1464 (44%)] ,Loss: 0.915691 ,Accuracy:85.35%, Train Mean IoU:0.37114\n",
      "Train Epoch: 93 [800/1464 (55%)] ,Loss: 0.673639 ,Accuracy:84.95%, Train Mean IoU:0.36977\n",
      "Train Epoch: 93 [960/1464 (66%)] ,Loss: 0.849157 ,Accuracy:85.22%, Train Mean IoU:0.37167\n",
      "Train Epoch: 93 [1120/1464 (77%)] ,Loss: 0.964127 ,Accuracy:85.05%, Train Mean IoU:0.36953\n",
      "Train Epoch: 93 [1280/1464 (87%)] ,Loss: 0.726405 ,Accuracy:84.75%, Train Mean IoU:0.36892\n",
      "Train Epoch: 93 [1440/1464 (98%)] ,Loss: 0.618990 ,Accuracy:84.88%, Train Mean IoU:0.36889\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 94 - - - - - - - -\n",
      "Train Epoch: 94 [160/1464 (11%)] ,Loss: 0.616341 ,Accuracy:84.10%, Train Mean IoU:0.36990\n",
      "Train Epoch: 94 [320/1464 (22%)] ,Loss: 0.772086 ,Accuracy:83.42%, Train Mean IoU:0.35768\n",
      "Train Epoch: 94 [480/1464 (33%)] ,Loss: 0.741698 ,Accuracy:83.65%, Train Mean IoU:0.36041\n",
      "Train Epoch: 94 [640/1464 (44%)] ,Loss: 0.683068 ,Accuracy:84.09%, Train Mean IoU:0.36097\n",
      "Train Epoch: 94 [800/1464 (55%)] ,Loss: 0.559800 ,Accuracy:84.32%, Train Mean IoU:0.36448\n",
      "Train Epoch: 94 [960/1464 (66%)] ,Loss: 1.157253 ,Accuracy:84.59%, Train Mean IoU:0.36523\n",
      "Train Epoch: 94 [1120/1464 (77%)] ,Loss: 0.739221 ,Accuracy:84.65%, Train Mean IoU:0.36663\n",
      "Train Epoch: 94 [1280/1464 (87%)] ,Loss: 0.870866 ,Accuracy:84.81%, Train Mean IoU:0.36712\n",
      "Train Epoch: 94 [1440/1464 (98%)] ,Loss: 1.026453 ,Accuracy:84.90%, Train Mean IoU:0.36917\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 95 - - - - - - - -\n",
      "Train Epoch: 95 [160/1464 (11%)] ,Loss: 0.748055 ,Accuracy:84.36%, Train Mean IoU:0.37082\n",
      "Train Epoch: 95 [320/1464 (22%)] ,Loss: 0.954856 ,Accuracy:84.49%, Train Mean IoU:0.37346\n",
      "Train Epoch: 95 [480/1464 (33%)] ,Loss: 0.761045 ,Accuracy:85.03%, Train Mean IoU:0.37607\n",
      "Train Epoch: 95 [640/1464 (44%)] ,Loss: 0.710906 ,Accuracy:84.86%, Train Mean IoU:0.37197\n",
      "Train Epoch: 95 [800/1464 (55%)] ,Loss: 1.410343 ,Accuracy:84.66%, Train Mean IoU:0.36986\n",
      "Train Epoch: 95 [960/1464 (66%)] ,Loss: 0.931962 ,Accuracy:84.80%, Train Mean IoU:0.36930\n",
      "Train Epoch: 95 [1120/1464 (77%)] ,Loss: 0.689509 ,Accuracy:84.77%, Train Mean IoU:0.36811\n",
      "Train Epoch: 95 [1280/1464 (87%)] ,Loss: 0.924267 ,Accuracy:84.71%, Train Mean IoU:0.36807\n",
      "Train Epoch: 95 [1440/1464 (98%)] ,Loss: 0.515734 ,Accuracy:84.87%, Train Mean IoU:0.36851\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 96 - - - - - - - -\n",
      "Train Epoch: 96 [160/1464 (11%)] ,Loss: 0.934924 ,Accuracy:85.43%, Train Mean IoU:0.37169\n",
      "Train Epoch: 96 [320/1464 (22%)] ,Loss: 1.044652 ,Accuracy:85.18%, Train Mean IoU:0.36884\n",
      "Train Epoch: 96 [480/1464 (33%)] ,Loss: 0.783560 ,Accuracy:85.26%, Train Mean IoU:0.37394\n",
      "Train Epoch: 96 [640/1464 (44%)] ,Loss: 1.194813 ,Accuracy:84.63%, Train Mean IoU:0.36657\n",
      "Train Epoch: 96 [800/1464 (55%)] ,Loss: 0.772370 ,Accuracy:85.00%, Train Mean IoU:0.36975\n",
      "Train Epoch: 96 [960/1464 (66%)] ,Loss: 0.734499 ,Accuracy:84.85%, Train Mean IoU:0.37029\n",
      "Train Epoch: 96 [1120/1464 (77%)] ,Loss: 0.849491 ,Accuracy:84.71%, Train Mean IoU:0.36858\n",
      "Train Epoch: 96 [1280/1464 (87%)] ,Loss: 0.818356 ,Accuracy:84.82%, Train Mean IoU:0.36987\n",
      "Train Epoch: 96 [1440/1464 (98%)] ,Loss: 0.816846 ,Accuracy:84.90%, Train Mean IoU:0.36905\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 97 - - - - - - - -\n",
      "Train Epoch: 97 [160/1464 (11%)] ,Loss: 0.850599 ,Accuracy:85.51%, Train Mean IoU:0.37455\n",
      "Train Epoch: 97 [320/1464 (22%)] ,Loss: 0.660042 ,Accuracy:85.67%, Train Mean IoU:0.37377\n",
      "Train Epoch: 97 [480/1464 (33%)] ,Loss: 0.869092 ,Accuracy:85.52%, Train Mean IoU:0.36937\n",
      "Train Epoch: 97 [640/1464 (44%)] ,Loss: 0.969389 ,Accuracy:85.22%, Train Mean IoU:0.36935\n",
      "Train Epoch: 97 [800/1464 (55%)] ,Loss: 1.034034 ,Accuracy:85.04%, Train Mean IoU:0.36719\n",
      "Train Epoch: 97 [960/1464 (66%)] ,Loss: 0.584259 ,Accuracy:85.08%, Train Mean IoU:0.36857\n",
      "Train Epoch: 97 [1120/1464 (77%)] ,Loss: 0.821254 ,Accuracy:84.92%, Train Mean IoU:0.36846\n",
      "Train Epoch: 97 [1280/1464 (87%)] ,Loss: 1.067142 ,Accuracy:84.75%, Train Mean IoU:0.36662\n",
      "Train Epoch: 97 [1440/1464 (98%)] ,Loss: 0.788504 ,Accuracy:84.87%, Train Mean IoU:0.36839\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 98 - - - - - - - -\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Epoch: 98 [160/1464 (11%)] ,Loss: 0.408519 ,Accuracy:85.65%, Train Mean IoU:0.36746\n",
      "Train Epoch: 98 [320/1464 (22%)] ,Loss: 0.726540 ,Accuracy:83.99%, Train Mean IoU:0.36116\n",
      "Train Epoch: 98 [480/1464 (33%)] ,Loss: 0.613069 ,Accuracy:84.45%, Train Mean IoU:0.36503\n",
      "Train Epoch: 98 [640/1464 (44%)] ,Loss: 0.693093 ,Accuracy:85.19%, Train Mean IoU:0.36771\n",
      "Train Epoch: 98 [800/1464 (55%)] ,Loss: 1.146097 ,Accuracy:85.03%, Train Mean IoU:0.36580\n",
      "Train Epoch: 98 [960/1464 (66%)] ,Loss: 0.757297 ,Accuracy:85.12%, Train Mean IoU:0.36904\n",
      "Train Epoch: 98 [1120/1464 (77%)] ,Loss: 0.815357 ,Accuracy:84.94%, Train Mean IoU:0.37004\n",
      "Train Epoch: 98 [1280/1464 (87%)] ,Loss: 0.968653 ,Accuracy:84.86%, Train Mean IoU:0.36800\n",
      "Train Epoch: 98 [1440/1464 (98%)] ,Loss: 0.946159 ,Accuracy:84.83%, Train Mean IoU:0.36842\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 99 - - - - - - - -\n",
      "Train Epoch: 99 [160/1464 (11%)] ,Loss: 0.683885 ,Accuracy:85.59%, Train Mean IoU:0.37652\n",
      "Train Epoch: 99 [320/1464 (22%)] ,Loss: 0.868022 ,Accuracy:84.43%, Train Mean IoU:0.36997\n",
      "Train Epoch: 99 [480/1464 (33%)] ,Loss: 0.657116 ,Accuracy:84.61%, Train Mean IoU:0.36668\n",
      "Train Epoch: 99 [640/1464 (44%)] ,Loss: 1.120299 ,Accuracy:84.61%, Train Mean IoU:0.36675\n",
      "Train Epoch: 99 [800/1464 (55%)] ,Loss: 0.732970 ,Accuracy:84.59%, Train Mean IoU:0.36661\n",
      "Train Epoch: 99 [960/1464 (66%)] ,Loss: 0.586496 ,Accuracy:84.73%, Train Mean IoU:0.36819\n",
      "Train Epoch: 99 [1120/1464 (77%)] ,Loss: 0.772305 ,Accuracy:84.57%, Train Mean IoU:0.36685\n",
      "Train Epoch: 99 [1280/1464 (87%)] ,Loss: 0.903895 ,Accuracy:84.63%, Train Mean IoU:0.36734\n",
      "Train Epoch: 99 [1440/1464 (98%)] ,Loss: 0.894011 ,Accuracy:84.87%, Train Mean IoU:0.36879\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "- - - - - epoch 100 - - - - - - - -\n",
      "Train Epoch: 100 [160/1464 (11%)] ,Loss: 0.749376 ,Accuracy:85.83%, Train Mean IoU:0.37088\n",
      "Train Epoch: 100 [320/1464 (22%)] ,Loss: 0.781327 ,Accuracy:85.20%, Train Mean IoU:0.36908\n",
      "Train Epoch: 100 [480/1464 (33%)] ,Loss: 1.050429 ,Accuracy:85.39%, Train Mean IoU:0.36934\n",
      "Train Epoch: 100 [640/1464 (44%)] ,Loss: 0.790296 ,Accuracy:85.00%, Train Mean IoU:0.36641\n",
      "Train Epoch: 100 [800/1464 (55%)] ,Loss: 0.767987 ,Accuracy:84.97%, Train Mean IoU:0.36635\n",
      "Train Epoch: 100 [960/1464 (66%)] ,Loss: 0.638261 ,Accuracy:84.78%, Train Mean IoU:0.36663\n",
      "Train Epoch: 100 [1120/1464 (77%)] ,Loss: 0.915262 ,Accuracy:84.68%, Train Mean IoU:0.36580\n",
      "Train Epoch: 100 [1280/1464 (87%)] ,Loss: 0.865641 ,Accuracy:84.76%, Train Mean IoU:0.36715\n",
      "Train Epoch: 100 [1440/1464 (98%)] ,Loss: 0.584984 ,Accuracy:84.89%, Train Mean IoU:0.36848\n",
      "- - - - - test stage- - - - - - - -\n",
      "Test_accuracy: 6.76% , Test Mean IoU: 2.94190 \n",
      "Training complete in 239m 12s\n"
     ]
    }
   ],
   "source": [
    "#模型评估\n",
    "train_start = time.time()\n",
    "epoch_y, epochacc_y ,testacc_y = train_test(optimizer)\n",
    "train_stop = time.time()\n",
    "train_total = train_stop - train_start\n",
    "print('Training complete in {:.0f}m {:.0f}s'.format(train_total // 60, train_total % 60))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#模型保存\n",
    "torch.save(model.state_dict(), './FCN_params.pkl')  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAmMAAAI1CAYAAACaH/2XAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSMKYEEIIIUQHSRgTQgghhOig/wINGiFELtZN9QAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f5bbdb1b2e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# 预测结果可视化处理\n",
    "cm = np.array(colormap).astype('uint8')\n",
    "def predict(im, label): # 预测结果\n",
    "    im = Variable(im.unsqueeze(0))\n",
    "    if use_gpu:\n",
    "        im = im.cuda()\n",
    "    out = model(im)\n",
    "    pred = out.max(1)[1].squeeze().cpu().data.numpy()\n",
    "    pred = cm[pred]\n",
    "    return pred, cm[label.numpy()]\n",
    "_, figs = plt.subplots(5, 3, figsize=(12, 10))\n",
    "for i in range(5):\n",
    "    test_data, test_label = testdata[i]\n",
    "    pred, label = predict(test_data, test_label)\n",
    "    figs[i, 0].imshow(Image.open(testdata.data_list[i]))\n",
    "    figs[i, 0].axes.get_xaxis().set_visible(False)\n",
    "    figs[i, 0].axes.get_yaxis().set_visible(False)\n",
    "    figs[i, 1].imshow(label)\n",
    "    figs[i, 1].axes.get_xaxis().set_visible(False)\n",
    "    figs[i, 1].axes.get_yaxis().set_visible(False)\n",
    "    figs[i, 2].imshow(pred)\n",
    "    figs[i, 2].axes.get_xaxis().set_visible(False)\n",
    "    figs[i, 2].axes.get_yaxis().set_visible(False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "结论:似乎有一丢丢用啊,尝试多几次训练可以观察效果会不会更好"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}