restructured project

README.md

@@ -1,20 +1,43 @@
-# DaKanji-ML
-This repository contains all machine learning models used in DaKanji. <br/>
-For every model there is a separate folder.
+# DaKanji Single Kanji CNN
 
-|      folder name |                     description                     |
-| :--------------: | :-------------------------------------------------: |
-| single_kanji_cnn | CNN which can detect single hand written characters |
+This CNN can recognize a single character (Kanji, Hiragana, Katakana).
+A list of all supported characters can be found [here](./labels.txt).
 
-If you are interested in using one of the models visit their folder.
-Every folder has more details on the model itself.
+## Training
+
+To generate the data necessary to train this CNN, the [single_kanji_data_gen notebook](single_kanji_data_gen.ipynb) is used.
+The training can than be done with the [single_kanji_cnn_training notebook](single_kanji_cnn_training.ipynb).
+
+## Inference
+
+In the releases section pretrained model weights can be found. Also a TensorFlow lite model is available.
+
+**Input:**
+The input should be a *grayscale, 8-bit* image of *any scale*.
+
+**Output:**
+A one-hot-vector containing the class probabilities (lines up with `labels.txt`).
 
 ## Setup development environment
+
 install all dependencies:
-```
+
+``` python
 python -m pip install wheel
 python -m pip install -r requirements.txt
 ```
+
 Now you should follow model specific setup steps.
 For this look at the README for the model you are interested in.
 
+## Credits
+
+The data on which the neural network was trained on was kindly provided by [ETL Character Database](http://etlcdb.db.aist.go.jp/obtaining-etl-character-database)
+
+[The KanjiVG dataset](https://kanjivg.tagaini.net/) was used to generate "handwritten" kanjis
+Papers:
+
+* [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](https://arxiv.org/abs/1905.11946)
+* [Recognizing Handwritten Japanese Characters Using Deep Convolutional Neural Networks](http://cs231n.stanford.edu/reports/2016/pdfs/262_Report.pdf)
+* [A neural framework for online recognition of handwritten Kanji characters](https://www.researchgate.net/publication/327893142_A_neural_framework_for_online_recognition_of_handwritten_Kanji_characters)
+* [Online Handwritten Kanji Recognition Based on Inter-stroke Grammar](https://www.researchgate.net/publication/4288187_Online_Handwritten_Kanji_Recognition_Based_on_Inter-stroke_Grammar)