-
Notifications
You must be signed in to change notification settings - Fork 0
/
data.py
40 lines (32 loc) · 1.75 KB
/
data.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
import os
import numpy as np
from imageio import imread
from skimage.transform import resize
from multiprocessing.pool import Pool
from tqdm import tqdm
def load_image(params):
path, size = params
return resize(imread(path), size).astype(np.float32)
def get_image_paths_by_directories(root_path, min_samples, target_size, supported_extensions=[".jpg", ".jpeg", ".png", ".gif", ".bmp"]):
"""Returns a list of images grouped by label. Takes the image directory as its label.
Works with nested directories."""
image_paths_by_label = []
with Pool() as pool:
for root, _, filenames in tqdm(os.walk(root_path)):
matching_files = [name for name in filenames if os.path.splitext(name)[1].lower() in supported_extensions]
if len(matching_files) >= min_samples:
image_paths_by_label.append([os.path.join(root, path) for path in matching_files])
return image_paths_by_label
def load_images_by_directories(root_path, min_samples, target_size, supported_extensions=[".jpg", ".jpeg", ".png", ".gif", ".bmp"]):
"""Returns a list of images grouped by label. Takes the image directory as its label.
Works with nested directories."""
images_by_label = []
with Pool() as pool:
for root, _, filenames in tqdm(os.walk(root_path)):
matching_files = [name for name in filenames if os.path.splitext(name)[1].lower() in supported_extensions]
if len(matching_files) >= min_samples:
images = pool.map(load_image, [(os.path.join(root, path), target_size) for path in matching_files])
images_by_label.append(np.array(images))
#if len(images_by_label) >= 8:
# break
return images_by_label