Submitted by Changjiang Cai.

Submission data

Full name	CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation
Description	We generate a matching cost volume leveraging both data with ground truth and conventional wisdom. We accomplish this by coalescing diverse evidence from a bidirectional matching process via random forest classifiers. We show that the resulting matching volume estimation method achieves similar accuracy to purely data-driven alternatives on benchmarks and that it generalizes to unseen data much better. For more details, please check our paper.
Parameters	* pi1 = 2.4 * pi2 = 24.25 * sgm_q1=3.3 * sgm_q2=1.4 * alpha=2.9 * tau_so=0.16 (There is a detailed setup about the parameters in our paper)
Publication title	CBMV: A Coalesced Bidirectional Matching Volume for Disparity Estimation
Publication authors	Konstantinos Batsos, Changjiang Cai, Philippos Mordohai
Publication venue	Computer Vision and Pattern Recognition (CVPR) 2018
Publication URL	https://arxiv.org/abs/1804.01967
Programming language(s)	Python,C/C++, with CUDA
Hardware	6 cores @ 3.0 Ghz, Nvidia TitanX
Website	https://github.com/kbatsos/CBMV
Source code or download URL	https://github.com/kbatsos/CBMV
Submission creation date	13 Jan, 2018
Last edited	10 Apr, 2018

High-res multi-view results

Low-res many-view results

Low-res two-view results

SLAM results