AAAI2020

Chained Representation Cycling: Learning to Estimate 3D Human Pose and Shape by Cycling Between Representations

Extending cycle consistency models to learn mappings between 2D images and parameterized 3D models.