| language: | |
| - en | |
| license: mit | |
| pipeline_tag: other | |
| tags: | |
| - PointCloud | |
| - PointCloudAnalysis | |
| - PointCloudSegmentation | |
| # PointCRA | |
| This is the official code repository for **PointCRA**, a point cloud analysis network proposed in our paper: | |
| > **Channel-Level Relation to Attentive Aggregation with Neighborhood-Homogeneity Constraint for Point Cloud Analysis** [[Arxiv](https://arxiv.org/abs/2605.02357)] [[GitHub](https://github.com/AGENT9717/PointCRA)] | |
| PointCRA introduces a novel network with a channel-level metric-based enhancement mechanism. Its core idea is to introduce temporal trend variation as a new evaluation dimension to avoid information loss caused by weight dimension collapse in existing spatial and channel attention mechanisms. The method leverages neighborhood homogeneity for weight calibration, offering an interpretable and efficient framework for tasks like 3D segmentation and classification. | |
| ## Citation | |
| If you find this work useful, please cite our paper: | |
| ```bibtex | |
| @article{shi2025pointcra, | |
| title = {Channel-Level Relation to Attentive Aggregation with Neighborhood-Homogeneity Constraint for Point Cloud Analysis}, | |
| author = {Jiaqi Shi, Jin Xiao, Xiaoguang Hu, Wenxuan Ji, Zichong Jia, Zifan Long, and Tianyou Chen}, | |
| journal = {arXiv preprint}, | |
| year = {2026} | |
| } | |
| ``` |