python - tf.image.pad_to_bounding_box VS tf.pad and tf.image.crop_to_bounding_box VS tf.slice -

i'd understand why 2 functions tf.image.crop_to_bounding_box , tf.image.pad_to_bounding_box exists, since behaviour of these 2 functions can done respectively tf.slice , tf.pad.

they not easier understand, , scope narrow since accept 3d , 4d tensors. furthermore, tend slower in terms of time of execution.

is there miss here ?

mostly use them tf.image.* easiness of use.

both crop_to_bounding_box , pad_to_bounding_box use slice , padunderneath, add checkings , constraints make sure don't spend hours trying debug slice/pad indices , offsets.