arXiv reaDer
Three-Filters-to-Normal+: Revisiting Discontinuity Discrimination in Depth-to-Normal Translation
This article introduces three-filters-to-normal+ (3F2N+), an extension of our previous work three-filters-to-normal (3F2N), with a specific focus on incorporating discontinuity discrimination capability into surface normal estimators (SNEs). 3F2N+ achieves this capability by utilizing a novel discontinuity discrimination module (DDM), which combines depth curvature minimization and correlation coefficient maximization through conditional random fields (CRFs). To evaluate the robustness of SNEs on noisy data, we create a large-scale synthetic surface normal (SSN) dataset containing 20 scenarios (ten indoor scenarios and ten outdoor scenarios with and without random Gaussian noise added to depth images). Extensive experiments demonstrate that 3F2N+ achieves greater performance than all other geometry-based surface normal estimators, with average angular errors of 7.85^∘, 8.95^∘, 9.25^∘, and 11.98^∘ on the clean-indoor, clean-outdoor, noisy-indoor, and noisy-outdoor datasets, respectively. We conduct three additional experiments to demonstrate the effectiveness of incorporating our proposed 3F2N+ into downstream robot perception tasks, including freespace detection, 6D object pose estimation, and point cloud completion. Our source code and datasets are publicly available at https://mias.group/3F2Nplus.
updated: Wed Dec 13 2023 08:18:49 GMT+0000 (UTC)
published: Wed Dec 13 2023 08:18:49 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト