Improve 3D object segmentation with scene labels by using superpoints with WHCN
Improve 3D object segmentation with scene labels by using superpoints with WHCN
Hypergraph Convolutional Network based Weakly Supervised Point Cloud Semantic Segmentation with Scene-Level Annotations
arXiv paper abstract https://arxiv.org/abs/2211.01174
arXiv PDF paper https://arxiv.org/pdf/2211.01174.pdf
Point cloud segmentation with scene-level annotations is a promising but challenging task.
… most popular … employ the class activation map (CAM) … However, these methods always suffer from the point imbalance among categories, as well as the sparse and incomplete supervision from CAM.
… propose a novel weighted hypergraph convolutional network-based method, called WHCN, to confront the challenges of learning point-wise labels from scene-level annotations.
Firstly … superpoints of a training point cloud are generated by exploiting the geometrically homogeneous partition.
… hypergraph is constructed … on … superpoint-level seeds … from … annotations … takes the hypergraph … learns to predict high-precision point-level pseudo labels by label propagation.
… proposed WHCN is effective to predict the point labels with scene annotations, and yields state-of-the-art results in the community …
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
