GANOCS: Domain Adaptation of Normalized Object Coordinate Prediction Using Generative Adversarial Training

Bahaa Aldeeb, Sahith Reddy Chada, Karthik Desingh

Keywords: Domain Adaptation, 2D-3D Correspondences, 3D Perception, Semi-supervised Learning

Abstract: Estimating 2D-3D correspondences has proven to be a very useful tool for category-level pose and scale estimation and robot manipulation tasks; however, it is hindered by the difficulty of obtaining 3D object models and labels. Simulation reduces the burden of labeling but introduces a gap between the training and operational domains. We introduce a novel architecture for integrating cross-domain data in the training of NOCS predictors (a form of 2D-3D correspondences). We leverage Generative-Adversarial-Networks (GANs) to avoid the need for burdensome real-data labeling by using a domain-agnostic discriminator as a supervisor. This work presents results demonstrating the potential of our method.

Submission Number: 22