I have a deep learning task involving geometry (keypoints/bounding boxes from medical images as input, in order to perform whole-image classification of a patient outcome po
