Pytorch-yolov3: OSError: image file is truncated (7 bytes not processed)

[ghimiredhikura]

It seems there is still some problem with Image loader.

 File "...\PyTorch-YOLOv3\utils\datasets.py", line 86, in __getitem__
    img = transforms.ToTensor()(Image.open(img_path).convert('RGB'))
  File "C:\ProgramData\Anaconda3\envs\ghimire\lib\site-packages\PIL\Image.py", line 899, in convert
    self.load()
  File "C:\ProgramData\Anaconda3\envs\ghimire\lib\site-packages\PIL\ImageFile.py", line 236, in load
    len(b))
OSError: image file is truncated (7 bytes not processed)

Therefore i'm using cv2 image loader.

[chuong]

It's probably you installed Pillow instead of PIL. Try adding these to datasets.py file at line 14:

from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

[ghimiredhikura]

It's probably you installed Pillow instead of PIL. Try adding these to datasets.py file at line 14:

from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

@chuong , Solved

Thank you

[Chee-11]

The method about:

from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

is ineffective to me, so I change the code:
img = transforms.ToTensor()(Image.open(img_path).convert('RGB'))
to this:
img = transforms.ToTensor()(cv2.cvtColor(cv2.imread(img_path), cv2.COLOR_BGR2RGB))
and it works.

[yosunpeng]

Thinks!