updates

data/get_coco_dataset.sh

@@ -36,4 +36,4 @@ paste <(awk "{print \"$PWD\"}" <trainvalno5k.part) trainvalno5k.part | tr -d '\t
 # sudo rm -rf train_images/._*
 # lastly convert each .tif to a .bmp for faster loading in cv2
 
-# /home/glenn_jocher3/coco/images/train2014/COCO_train2014_000000167126.jpg  # bad image??
+# ./coco/images/train2014/COCO_train2014_000000167126.jpg  # corrupted image

requirements.txt

@@ -1,5 +1,7 @@
 # pip3 install -U -r requirements.txt
-# conda install numpy opencv matplotlib tqdm pillow && conda install pytorch torchvision -c pytorch
+# conda install numpy opencv matplotlib tqdm pillow
+# conda install pytorch torchvision -c pytorch
+# conda install -c conda-forge scikit-image
 numpy
 opencv-python
 torch >= 1.1.0

utils/datasets.py

@@ -223,6 +223,16 @@ class LoadImagesAndLabels(Dataset):  # for training/testing
                     pass  # print('Warning: missing labels for %s' % self.img_files[i])  # missing label file
             assert len(np.concatenate(self.labels, 0)) > 0, 'No labels found. Incorrect label paths provided.'
 
+        # Detect corrupted images https://medium.com/joelthchao/programmatically-detect-corrupted-image-8c1b2006c3d3
+        detect_corrupted_images = False
+        if detect_corrupted_images:
+            from skimage import io  # conda install -c conda-forge scikit-image
+            for file in tqdm(self.img_files, desc='Detecting corrupted images'):
+                try:
+                    _ = io.imread(file)
+                except:
+                    print('Corrupted image detected: %s' % file)
+
     def __len__(self):
         return len(self.img_files)