One-Inlier is First: Towards Efficient Position Encoding for Point Cloud Registration

Authors: Fan Yang, Lin Guo, Zhi Chen, Wenbing Tao

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We evaluate our approach on indoor 3DMatch [49] and 3DLo Match [18] benchmarks (Sec. 4.1), and outdoor KITTI odometry [14] benchmark (Sec. 4.2). The point cloud pairs in 3DMatch have > 30% overlap, while those in 3DLo Match have low overlap of 10% 30%. KITTI odometry is an outdoor driving scenario sparse point cloud dataset acquired by Li DAR.
Researcher Affiliation Academia Fan Yang Lin Guo Zhi Chen Wenbing Tao School of Artificial Intelligence and Automation Huazhong University of Science and Technology, Wuhan 430074, China {fanyang,linguo,z_chen,wenbingtao}@hust.edu.cn
Pseudocode No The paper illustrates the proposed method with diagrams (e.g., Figure 1, Figure 2, Figure 4) and describes its steps in the text. However, it does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks, nor does it present structured steps formatted like code.
Open Source Code Yes 3. If you ran experiments...(a) Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes] See the supplemental material.
Open Datasets Yes We evaluate our approach on indoor 3DMatch [49] and 3DLo Match [18] benchmarks (Sec. 4.1), and outdoor KITTI odometry [14] benchmark (Sec. 4.2).
Dataset Splits No Details about the datasets and implementation are provided in the supplementary material.
Hardware Specification No 3. If you ran experiments...(d) Did you include the total amount of compute and the type of resources used (e.g., type of GPUs, internal cluster, or cloud provider)? [Yes] See the supplemental material.
Software Dependencies No The paper mentions several software components and algorithms such as 'KPConv [38]', 'differentiable optimal transport layer [31]', 'cross-attention [41, 31]', 'self-attention', 'MLP', 'weighted SVD [5]', and 'RANSAC50k'. However, it does not provide specific version numbers for these software dependencies.
Experiment Setup Yes Following [35, 48, 36], our method employs a coarse-tofine manner to find correspondences. In the coarse stage, we first use the KPConv [38] to down-sample raw input points to uniformly distributed nodes X R M 3 and Y R N 3, and learn the associated features F X R M D and F Y R N D. Then coarse correspondences are generated by matching the nodes.