-
Notifications
You must be signed in to change notification settings - Fork 12
/
setup_database_GAIC.m
44 lines (37 loc) · 1.55 KB
/
setup_database_GAIC.m
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
function imdb = setup_database_GAIC(trainNum,seed)
if ~exist(['imdb_GAIC' num2str(trainNum) '.mat'],'file')
train_img_dir = fullfile('dataset','GAIC','images','train');
train_img_path = dir(train_img_dir);
train_img_path = fullfile(train_img_dir,{train_img_path(3:end).name});
test_img_dir = fullfile('dataset','GAIC','images','test');
test_img_path = dir(test_img_dir);
test_img_path = fullfile(test_img_dir,{test_img_path(3:end).name});
img_path = cat(2,train_img_path,test_img_path);
for i = 1:numel(img_path)
[~,imgname,~] = fileparts(img_path{i});
anno = load(fullfile('dataset','GAIC','annotations',[imgname '.txt']));
boxes{i} = anno(:,1:4)';
avg_scores(i,:) = anno(:,5)';
kill = (avg_scores(i,:)==-2);
gt_scores{i} = avg_scores(i,~kill);
boxes{i} = boxes{i}(:,~kill);
end
rng(seed)
trainSet = randperm(numel(train_img_path),trainNum);
valSet = setdiff(1:numel(train_img_path),trainSet);
testSet = numel(train_img_path)+1:numel(train_img_path)+numel(test_img_path);
set = ones(1,numel(img_path));
set(valSet) = 3;
set(testSet) = 2;
imdb.images.trainSet = trainSet;
imdb.images.valSet = valSet;
imdb.images.testSet = testSet;
imdb.bbox.boxes = boxes;
imdb.bbox.gt_scores = gt_scores;
imdb.images.set = set;
imdb.meta.sets = {'train', 'val', 'test'} ;
imdb.meta.img_path = img_path;
save(['imdb_GAIC' num2str(trainNum) '.mat'],'imdb');
else
load(['imdb_GAIC' num2str(trainNum) '.mat'],'imdb');
end