The proposed layers lead toamore data efficient representation and areduction in computation by exploiting symmetry. We evaluate chirality nets on the task ofhuman poseregression, which naturally exploits theleft/right mirroring ofthe human body.

Compared toprevious works, our approach can learn landmarks that are more flexible in terms of capturing large changes in viewpoint.

However, developing models that can process weight-space objects is challenging due to their high dimensional nature.

A truly generalizable approach to rigid segmentation and motion estimation is fundamental to 3D understanding of articulated objects and moving scenes.

Any explanation that faithfully explains this type of model needs to be in agreement with this invariance property.