dgs.utils.validation.validate_image

dgs.utils.validation.validate_image(images: torchvision.tv_tensors.Image | torch.Tensor | torch.Tensor, length: int | None = None, dims: int | None = 4) → torchvision.tv_tensors.Image | torch.Tensor[source]

Given one single image or a stacked batch images, validate them and return a torchvision-tensor image.

Parameters:

  • images – torch tensor or tv_tensors.Image object

  • length – The number of items or batch-size the tensor should have. Default None does not validate the length.

  • dims – Number of dimensions img should have. Use None to not force any number of dimensions. Defaults to four dimensions with the image dimensions as [B x C x H x W].

Returns:

The images as tv_tensors.Image object with exactly dims dimensions.

Return type:

Image

Raises:

  • TypeError – If images is not a Tensor or cannot be cast to one.

  • ValueError – If the dimension of the images channels is wrong.