Manipulation

Equivariant Single View Pose Prediction Via Induced and Restriction Representations

We train an equivariant network for pose prediction from single 2D image by using induced and restricted representations.