GitLab

Updating transform_input_data to resize original image. This is necessary for result_dict_for_single_example(), since it expects the input image and groundtruth masks to be of the same spatial dimensions.

PiperOrigin-RevId: 189786443