Semantic Correspondence via 2D-3D-2D Cycle

TL;DR

This paper introduces a novel approach for semantic correspondence that leverages 3D information to improve accuracy in 2D image matching, explicitly reasoning about occlusion and visibility.

Contribution

It proposes a 3D-aware method for semantic correspondence that outperforms previous 2D-only approaches by explicitly modeling 3D object properties.

Findings

Achieves superior results on standard semantic benchmarks.

Effectively reasons about object occlusion and visibility.

Provides detailed analysis of network components.

Abstract

Visual semantic correspondence is an important topic in computer vision and could help machine understand objects in our daily life. However, most previous methods directly train on correspondences in 2D images, which is end-to-end but loses plenty of information in 3D spaces. In this paper, we propose a new method on predicting semantic correspondences by leveraging it to 3D domain and then project corresponding 3D models back to 2D domain, with their semantic labels. Our method leverages the advantages in 3D vision and can explicitly reason about objects self-occlusion and visibility. We show that our method gives comparative and even superior results on standard semantic benchmarks. We also conduct thorough and detailed experiments to analyze our network components. The code and experiments are publicly available at https://github.com/qq456cvb/SemanticTransfer.