Adding tiny_imagenet clean up code

modules/base_utils/fix_val.py

@@ -0,0 +1,36 @@
+import os
+
+DATA_DIR = 'data/tiny-imagenet-200/'
+VALID_DIR = DATA_DIR + 'val'
+
+# Create separate validation subfolders for the validation images based on
+# their labels indicated in the val_annotations txt file
+val_img_dir = os.path.join(VALID_DIR, 'images')
+fp = open(os.path.join(VALID_DIR, 'val_annotations.txt'), 'r')
+data = fp.readlines()
+
+# Create dictionary to store img filename (word 0) and corresponding
+# label (word 1) for every line in the txt file (as key value pair)
+val_img_dict = {}
+for line in data:
+ words = line.split('\t')
+ val_img_dict[words[0]] = words[1]
+fp.close()
+
+# Create subfolders (if not present) for validation images based on label ,
+# and move images into the respective folders
+for img, folder in val_img_dict.items():
+ newpath = (os.path.join(val_img_dir, folder))
+ if not os.path.exists(newpath):
+ os.makedirs(newpath)
+ if os.path.exists(os.path.join(val_img_dir, img)):
+ os.rename(os.path.join(val_img_dir, img), os.path.join(newpath, img))
+
+# Save class names (for corresponding labels) as dict from words.txt file
+class_to_name_dict = dict()
+fp = open(os.path.join(DATA_DIR, 'words.txt'), 'r')
+data = fp.readlines()
+for line in data:
+ words = line.strip('\n').split('\t')
+ class_to_name_dict[words[0]] = words[1].split(',')[0]
+fp.close()

modules/base_utils/tiny_imagenet_setup.sh

@@ -1,2 +1,3 @@
 wget -nc http:https://cs231n.stanford.edu/tiny-imagenet-200.zip
-unzip tiny-imagenet-200.zip -d data/
+unzip tiny-imagenet-200.zip -d data/
+python modules/base_utils/fix_val.py