what dimension is the depth channel?

[whisney]

volumentations/volumentations/augmentations/functional.py

Line 309 in 50a1193

tmp = img[:, :, d] # [H, W, C]

Is there a mistake in the dimension of depth here?

[ZFTurbo]

Why do you think it's mistake? Any of first 3 dimensions can be depth. In this code it's assumed that 3rd dimension is depth.

[whisney]

But the line 259 and line 281 in the same function, it's assumed that 1st dimension is depth.

volumentations/volumentations/augmentations/functional.py

Line 259 in 50a1193

depth, height, width = img.shape[:3]

volumentations/volumentations/augmentations/functional.py

Line 281 in 50a1193

tmp = img[d, :, :] # [D, H, W, C]

[ZFTurbo]

I think it's not really a bug, but 2 different dimensions used for transforms. But need to investigate. Thanks.