用分层高度的2D鸟瞰图实现毫秒级行人检测，适合实时机器人部署。
TriBand-BEV: Real-Time LiDAR-Only 3D Pedestrian Detection via Height-Aware BEV and High-Resolution Feature Fusion
- 将3D点云分三高度带转为轻量2D鸟瞰图，降维后做2D检测
- 在KITTI上达49帧/秒，行人检测平均精度最高58.7%
- 抗遮挡稳定，代码开源，适合嵌入式实时系统
安全的自动驾驶与移动机器人需要快速的实时3D感知，尤其针对行人等弱势道路使用者。本文提出一种新型鸟瞰图（BEV）编码方式，将完整3D LiDAR点云映射为三个高度带组成的轻量2D BEV张量，将3D检测转化为2D检测问题，并从BEV输出重建3D框。单网络可一次性检测车辆、行人和骑车人。骨干采用深层区域注意力，从P1到P4的分层双向颈部融合上下文与细节信息，检测头使用分布焦点学习预测侧向偏移，结合旋转IoU损失。训练时引入小范围垂直重采样与温和的通道反射率抖动以防止过拟合。3D重建阶段使用四分位距（IQR）滤波去除噪声与离群点。在KITTI数据集上，TriBand-BEV在单个消费级显卡上实现49 FPS，行人BEV AP分别达到58.7%（易）、52.6%（中）、47.2%（难），超越Complex-YOLO，提升+12.6%、+7.5%、+3.1%。定性结果显示其在遮挡下仍具稳定性。整体流程紧凑，已准备就绪用于实时机器人部署。源码已公开于GitHub。
原文摘要 · Abstract (English)
Safe autonomous agents and mobile robots need fast real time 3D perception, especially for vulnerable road users (VRUs) such as pedestrians. We introduce a new bird's eye view (BEV) encoding, which maps the full 3D LiDAR point cloud into a light-weight 2D BEV tensor with three height bands. We explicitly reformulate 3D detection as a 2D detection problem and then reconstruct 3D boxes from the BEV outputs. A single network detects cars, pedestrians, and cyclists in one pass. The backbone uses area attention at deep stages, a hierarchical bidirectional neck over P1 to P4 fuses context and detail, and the head predicts oriented boxes with distribution focal learning for side offsets and a rotated IoU loss. Training applies a small vertical re bin and a mild reflectance jitter in channel space to resist memorization. We use an interquartile range (IQR) filter to remove noisy and outlier LiDAR points during 3D reconstruction. On KITTI dataset, TriBand-BEV attains 58.7/52.6/47.2 pedestrian BEV AP(%) for easy, moderate, and hard at 49 FPS on a single consumer GPU, surpassing Complex-YOLO, with gains of +12.6%, +7.5%, and +3.1%. Qualitative scenes show stable detection under occlusion. The pipeline is compact and ready for real time robotic deployment. Our source code is publicly available on GitHub.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。