C3W2A_Assignment Center Cropping

[Nydia]

@Vasili_Reikh

    import torchvision.transforms as transforms 
    transform = transforms.CenterCrop((new_shape[-2], new_shape[-1])) 
    cropped_image = transform(image)

It actually passes the unit tests (with a ‘Passed’ message), but it cannot pass the grader (Cell #5. Can’t compile the student’s code. Error: TypeError(“‘builtin_function_or_method’ object is not iterable”,).
p.s. (tried with just one shape, as soon as the image is a square) - the same error with a grader, but interestingly, I can train the U-Net, but cannot submit the assignment)

The error is coming from passing directly from this function:

def crop(image, new_shape):

the image input:

@Vasili_Reikh

cropped_image = transform(image)

you should slice image before feeding it into transform.

A hint to to this: consider using slicing with negative indices to handle both even and odd dimensions more consistently. This can help simplify the calculations for starting and final indices, making the code more concise and less prone to errors related to even/odd dimensions.

[Wendy]

@Vasili_Reikh, the underlying problem with using CenterCrop() is due to a version mismatch, as mentioned in the first posts on this thread - although back then, the problem was happening in the assignment. Now the assignment works with CenterCrop(), but the grader is still using an older version of python or pytorch. I’ll ping the developers to remind them the grader needs to be updated for this.

In the meantime, instead of using CenterCrop(), you will need to do your own array-slicing to crop the image yourself.

There’s also some discussion of this issue on this thread, if you’re interested:

C3W2A_autograder_can't compile code

Yes, then that’s the problem. This is due to a version mismatch with the grader, in the case of Greg, he is using transform.functional.crop() and the thing with this and the CenterCrop function is that in the earlier versions the return type is “PIL Image” but in the higher versions the return type is “PIL Image or Tensors”. Our notebooks use a higher version of Torchvision and hence we don’t see any errors there but the grader might be using an earlier version of Torchvision, which leads to thi…