MoCha-V2's results in the original manuscript were
removed from the list, this is the result of a re-
upload version.MoCha-V2 introduces the Motif
Correlation Graph (MCG) to capture recurring
textures, which are referred to as “motifs”
within feature channels. These motifs reconstruct
geometric structures and are learned in a more
interpretable way. Subsequently, we integrate
features from multiple frequency domains through
wavelet inverse transformation. The resulting
motif features are utilized to restore geometric
structures in the stereo matching process.
|
@article{chen2024motif,
title={Motif Channel Opened in a White-Box:
Stereo Matching via Motif Correlation Graph},
author={Chen, Ziyang and Zhang, Yongjun and Li,
Wenting and Wang, Bingshu and Zhao, Yong and Chen,
CL},
journal={arXiv preprint arXiv:2411.12426},
year={2024}
}
@inproceedings{chen2024mocha,
title={MoCha-Stereo: Motif Channel Attention
Network for Stereo Matching},
author={Chen, Ziyang and Long, Wei and Yao, He
and Zhang, Yongjun and Wang, Bingshu and Qin,
Yongbin and Wu, Jia},
booktitle={Proceedings of the IEEE/CVF
Conference on Computer Vision and Pattern
Recognition},
pages={27768--27777},
year={2024}
} |