I am using an unofficial implementation of roi align for a project (link below). I came across a function called adaptive average pool2d and wondered if the same can be done in combination with torchvision crop.
Will they doing the same job or there’ll be a difference?
ROI Align unofficial implementation - https://github.com/longcw/RoIAlign.pytorch
Adaptive average pool - https://pytorch.org/docs/stable/nn.html#torch.nn.functional.adaptive_avg_pool2d
Torchvision crop - https://pytorch.org/docs/stable/torchvision/transforms.html#torchvision.transforms.functional.crop