Torchvision functional transforms for transforming pixel locations

Suppose I want to perform detect task and I need bounding box coordinates for the regressor in the given Image. Now if I rotate the image or pad it, that pixel would have shifted to some new location.
Is there an easy or some inbuilt way that would help me convert the pixel coordinates in original image to pixel coordinates in new image.

I can obviously write the functions but is messy work and inbuilt functionality would be better.
Thanks for the help