LiDAR-based 3D point cloud recognition has benefited various applications. Without specially considering the LiDAR point distribution, most current methods suffer from information disconnection and limited receptive field, especially for the sparse distant points. In this work, we study the varying-sparsity distribution of LiDAR points and present SphereFormer to directly aggregate information from dense close points to the sparse distant ones. We design radial window self-attention that partitions the space into multiple non-overlapping narrow and long windows. It overcomes the disconnection issue and enlarges the receptive field smoothly and dramatically, which significantly boosts the performance of sparse distant points. Moreover, to fit the narrow and long windows, we propose exponential splitting to yield fine-grained position encoding and dynamic feature selection to increase model representation ability. Notably, our method ranks 1st on both nuScenes and SemanticKITTI semantic segmentation benchmarks with 81.9% and 74.8% mIoU, respectively. Also, we achieve the 3rd place on nuScenes object detection benchmark with 72.8% NDS and 68.5% mAP. Code is available at https://github.com/dvlab-research/SphereFormer.git.

翻译：基于LiDAR的三维点云识别已惠及多种应用。由于未特别考虑LiDAR点云分布特性，当前多数方法存在信息割裂与感受野受限的问题，尤其对于稀疏远距离点云。本文研究了LiDAR点云疏密变化的分布规律，提出SphereFormer方法，实现从密集近点直接向稀疏远点进行信息聚合。我们设计了径向窗口自注意力机制，将空间划分为若干非重叠的窄长窗口。该方法克服了信息割裂问题，并平滑且显著地扩大了感受野，大幅提升稀疏远距离点云的性能表现。此外，为适配窄长窗口结构，我们提出指数分割方法以生成细粒度位置编码，并采用动态特征选择增强模型表征能力。值得关注的是，本方法在nuScenes和SemanticKITTI语义分割基准测试中均位列第一，分别取得81.9%和74.8%的mIoU；在nuScenes目标检测基准测试中取得第三名，NDS达72.8%、mAP达68.5%。代码已开源：https://github.com/dvlab-research/SphereFormer.git