Skip to main content Skip to secondary navigation
Journal Article

DCL: Differential Contrastive Learning for Geometry-Aware Depth Synthesis

DCL results

We describe a method for realistic depth synthesis that learns diverse variations from the real depth scans and ensures geometric consistency for effective synthetic-to-real transfer. Unlike general image synthesis pipelines, where geometries are mostly ignored, we treat geometries carried by the depth based on their own existence. We propose differential contrastive learning that explicitly enforces the underlying geometric properties to be invariant regarding the real variations been learned. The resulting depth synthesis method is task-agnostic and can be used for training any task-specific networks with synthetic labels. We demonstrate the effectiveness of the proposed method by extensive evaluations on downstream real-world geometric reasoning tasks. We show our method achieves better synthetic-to-real transfer performance than the other state-of-the-art. When fine-tuned on a small number of real-world annotations, our method can even surpass the fully supervised baselines.

Paper    Code

Yanchao Yang
Yuefan Shen
Youzi Zheng
C. Karen Liu
Leonidas J. Guibas
Journal Name
IEEE International Conference on Robotics and Automation (RA-L and ICRA), 2022
Publication Date
May, 2022