https://github.com/pytorch/vision/blob/ca2212438fdd8ce29b66999ed70ed54b0f9372d1/torchvision/transforms/functional.py#L388C5-L388C17 Type hints are not correct, `pad` and `resize` support both tensors and PIL images