PV-RCNN++: semantical point-voxel feature interaction for 3D object detection

Wu, Peng; Gu, Lipeng; Yan, Xuefeng; Xie, Haoran; Wang, Fu Lee; Cheng, Gary; Wei, Mingqiang

doi:10.1007/s00371-022-02672-2

Cited by 16 publications

(1 citation statement)

References 56 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Point-voxel as input and voxelization operation have been widely used in point cloud detection, segmentation and classification [18,19,10,24,7,9,22,2,1], which shows the effectiveness and efficiency of voxelization. Our presented point-voxel encoder consists of three modules: local encoding, Transformer and pyramid feature interaction.…”

Section: Point-voxel Encodermentioning

confidence: 99%

Point-Voxel Adaptive Feature Abstraction for Robust Point Cloud Classification

Lifa¹,

Lin²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Great progress has been made in point cloud classification with learning-based methods. However, complex scene and sensor inaccuracy in real-world application make point cloud data suffer from corruptions, such as occlusion, noise and outliers. In this work, we propose Point-Voxel based Adaptive (PV-Ada) feature abstraction for robust point cloud classification under various corruptions. Specifically, the proposed framework iteratively voxelize the point cloud and extract pointvoxel feature with shared local encoding and Transformer. Then, adaptive max-pooling is proposed to robustly aggregate the point cloud feature for classification. Experiments on ModelNet-C dataset demonstrate that PV-Ada outperforms the state-of-the-art methods. In particular, we rank the 2 nd place in ModelNet-C classification track of PointCloud-C Challenge 2022, with Overall Accuracy (OA) being 0.865. Code will be available at https://github.com/zhulf0804/PV-Ada.

show abstract