But where capsules are concerned ... predicts the pose of already discovered parts of images (segmented by an autoencoder) using a mixture of pose predictions, and that models each image pixel ...