RoadOcc：学习何时持久化、传输或刷新记忆以进行路侧占用预测
RoadOcc Learns When to Persist, Transport, or Refresh Memory for Roadside Occupancy Prediction
浏览论文内容
中文总结 AI 辅助
RoadOcc通过软路由在固定坐标历史、速度寻址历史和当前证据间选择，结合DCA、VVE和VDSF，在InfraOcc上达到65.29 mIoU，显著提升动态占用预测性能。
中文摘要 AI 辅助
固定的路侧摄像头反复观察一个由稀疏移动交通叠加的稳定场景。时间记忆可以恢复弱观测，但在过时位置重用移动证据会破坏占用预测。运动补偿解决了位移问题，而对由此产生的历史的依赖仍然是一个独立的学习问题。我们提出了RoadOcc，它学习在固定坐标历史（持久化）、速度寻址历史（传输）和当前证据（刷新）之间进行软路由。运动状态和类别一致的历史支持监督这些来源选择。动态感知交叉注意力（DCA）更新候选位置，多尺度体素速度估计（VVE）从多尺度当前-历史对应构建传输地址，速度引导的动态稀疏融合（VDSF）在固定稀疏令牌预算下组合路由证据。在InfraOcc上，RoadOcc达到65.29 mIoU和32.37动态mIoU，比STCOcc分别提高4.44和4.71。受控地址实验表明，VVE比固定坐标读取将动态mIoU提高0.87。在三个种子上，监督的P/T/R比运动校正检索增加1.40动态点，而移除刷新则损失0.32点。来自两个迁移模型、Occ3D-nuScenes和更长间隔的结果提供了额外支持。代码将发布。
英文摘要
Fixed roadside cameras repeatedly observe a stable scene overlaid by sparse moving traffic. Temporal memory can recover weak observations, but reusing moving evidence at stale locations can corrupt occupancy predictions. Motion compensation addresses displacement, while reliance on the resulting history remains a separate learning problem. We introduce RoadOcc, which learns soft routing among fixed-coordinate history (\emph{Persist}), velocity-addressed history (\emph{Transport}), and current evidence (\emph{Refresh}). Motion state and class-consistent historical support supervise these source choices. Dynamic-aware cross-attention (DCA) updates candidate locations, multi-scale voxel velocity estimation (VVE) constructs transport addresses from multi-scale current--history correspondence, and velocity-guided dynamic sparse fusion (VDSF) combines routed evidence under fixed sparse-token budgets. On InfraOcc, RoadOcc reaches 65.29 mIoU and 32.37 dynamic mIoU, gains of 4.44 and 4.71 over STCOcc. Controlled address experiments show that VVE raises dynamic mIoU by 0.87 over fixed-coordinate reading. Across three seeds, supervised P/T/R adds 1.40 dynamic points over motion-corrected retrieval, while removing Refresh costs 0.32 points. Results from two transfer models, Occ3D-nuScenes, and longer intervals provide additional support. Code will be released.
发表机构
- College of Information Science and Electronic Engineering, Zhejiang University（浙江大学信息科学与电子工程学院）
- School of Mechanical and Aerospace Engineering, Nanyang Technological University（南洋理工大学机械与航空航天工程学院）
- School of Automotive Studies, Tongji University（同济大学汽车学院）
机构由 AI 辅助整理，请以论文原文为准。
