-
Notifications
You must be signed in to change notification settings - Fork 0
/
split_data_train_val.py
53 lines (43 loc) · 2.13 KB
/
split_data_train_val.py
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
44
45
46
47
48
49
50
51
52
53
import os
import random
def main():
val_pct = 0.2 # 20%
path_data = "C:/Users/kevin/git-workspace/tf-platelets/Data/"
path_train_data = "C:/Users/kevin/git-workspace/tf-platelets/Data/Training/"
path_val_data = "C:/Users/kevin/git-workspace/tf-platelets/Data/Validation/"
data_labels = os.listdir(path_data)
if not os.path.exists(path_train_data):
os.mkdir(path_train_data)
for label in data_labels:
if not os.path.exists(os.path.join(path_train_data, label)):
os.mkdir(os.path.join(path_train_data, label))
if not os.path.exists(path_val_data):
os.mkdir(path_val_data)
for label in data_labels:
if not os.path.exists(os.path.join(path_val_data, label)):
os.mkdir(os.path.join(path_val_data, label))
num_samples = {}
for j, label in enumerate(data_labels):
if label == "Training" or label == "Validation":
continue
num_samples[label] = len([1 for x in list(os.scandir(os.path.join(
path_data, label))) if x.is_file() and os.path.splitext(x)[1] == ".png"])
for j, label in enumerate(data_labels):
if label == "Training" or label == "Validation":
continue
# val_ind = np.random.choice(num_samples[label], int(num_samples[label]*val_pct))
val_ind = random.sample(range(num_samples[label]), int(num_samples[label]*val_pct))
val_ind.sort()
print(val_ind[0:10])
print(f'({num_samples[label]},{int(num_samples[label]*val_pct)})')
image_names = [x.name for x in list(os.scandir(os.path.join(
path_data, label))) if x.is_file() and os.path.splitext(x)[1] == ".png"]
for i, image_name in enumerate(image_names):
if val_ind and i == val_ind[0]:
val_ind.pop(0)
os.rename(os.path.join(path_data, label, image_name), os.path.join(path_val_data, label, image_name))
else:
os.rename(os.path.join(path_data, label, image_name), os.path.join(path_train_data, label, image_name))
os.rmdir(os.path.join(path_data, label))
if __name__ == "__main__":
main()