Search papers, labs, and topics across Lattice.
This paper introduces a hybrid self-supervised method for non-rigid 3D shape matching that employs a coarse-to-fine strategy to ensure consistency between initial mappings and refined correspondences. Utilizing a dual-branch architecture with Laplacian and elastic basis streams, the method achieves state-of-the-art performance while maintaining computational efficiency across various challenging scenarios, such as non-isometric deformations and topological noise. The authors also demonstrate that integrating contrastive energies enhances feature discrimination, leading to consistent improvements over existing methods.
State-of-the-art non-rigid 3D shape matching is achieved through a novel coarse-to-fine self-supervised approach that balances efficiency and accuracy.
Non-rigid 3D shape matching is a fundamental task in computer vision and graphics. In this paper, we propose a hybrid self-supervised method based on a coarse-to-fine strategy, which ensures consistency between the coarse mapping and the refined correspondence produced by our refinement module. The architecture features a dual-branch design, consisting of two symmetric functional map learning streams: one based on the Laplacian basis and the other utilizing the elastic basis. Extensive experiments show that our approach not only maintains computational efficiency, but also achieves state-of-the-art performance across a variety of challenging scenarios, including non-isometric deformations and topological noise. Finally, we rigorously demonstrate that contrastive energies promote feature discrimination. Furthermore, integrating these energies with existing methods yields consistent improvements, validating the overall efficacy of our approach. Our code is available at https://github.com/LuoFeifan77/Coarse-to-Fine-Hybrid-Self-Supervised-Matching.
