Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
The paper proposes Chirality Nets for human pose regression, where the resulting network layers are equivariant to chirality transformations. Overall the paper presents some significantly novel results in a well written and intuitive manner. The experimental evaluation convincingly supports the proposed approach. The reviewers and AC consistently agree that the submission is of significant interest and novelty, and that the authors feedback has adequately addressed the points raised in the reviews.