Real-time 3D reconstruction despite occlusion using motion prediction with OcclusionFusion
Real-time 3D reconstruction despite occlusion using motion prediction with OcclusionFusion
OcclusionFusion: Occlusion-aware Motion Estimation for Real-time Dynamic 3D Reconstruction
arXiv paper abstract https://arxiv.org/abs/2203.07977
arXiv PDF paper https://arxiv.org/pdf/2203.07977.pdf
Project page https://wenbin-lin.github.io/OcclusionFusion
RGBD-based real-time dynamic 3D reconstruction suffers from inaccurate inter-frame motion estimation as errors may accumulate with online tracking.
This problem is even more severe for single-view-based systems due to strong occlusions.
… propose OcclusionFusion, a novel method to calculate occlusion-aware 3D motion to guide the reconstruction.
… the motion of visible regions is first estimated and combined with temporal information to infer the motion of the occluded regions
… method computes the confidence of the estimated motion by modeling the network output with a probabilistic model, which alleviates untrustworthy motions and enables robust tracking.
… technique outperforms existing single-view-based real-time methods by a large margin. … can handle long and challenging motion sequences.
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
