发表机构
Nantes Université; University of Nottingham; University of Queensland; TU Delft（南特大学; 诺丁汉大学; 昆士兰大学; 代尔夫特理工大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对激光雷达场景流估计的单一模式问题，本文提出无训练几何框架CorrelationFlow，在Argoverse 2026挑战赛多域测试集上的无监督方法中排名第二，且远距离性能优于学习方法。
AI 中文摘要
激光雷达场景流估计已陷入单一模式：几乎所有近期方法都采用相同的前馈架构和同一类自监督损失，相互继承假设与盲点。当这些假设失效时（如针对稀疏、远距离或快速移动物体），所有基于该框架的方法都会同步失效，增加参数或模拟训练数据无法修正公式本身的缺陷。本文采用相反路径，提出CorrelationFlow，一种无训练的几何框架，将场景流简化为两个经典操作：鸟瞰图占据图像上的连通分量标记与相关性最大化。物体被隔离为时空连通分量，其运动通过相关性峰值恢复，所得速度传播至所有成员点。由于该密集相关性需评估每个聚类的所有候选位移并依赖过去帧，因此开发了稀疏版本，仅通过匹配边界关键点处的轻量占据描述符，即可在单帧对上运行。因无任何训练，故无需继承任何假设：在Argoverse 2 2026场景流挑战赛的多域测试集（涵盖5个异构传感器与平台的数据集）上，CorrelationFlow在无监督方法中排名第二，且在远距离（学习方法的共享假设在此失效）时性能下降最平缓。结果表明，场景流问题的很大一部分可通过经典计算机视觉解决，进展可能需要质疑公式本身，而非对其进行扩展。
英文摘要
LiDAR scene flow estimation has settled into a monoculture: nearly all recent methods share the same feed-forward architecture and the same family of self-supervised losses, inheriting each other's assumptions, and each other's blind spots. When those assumptions fail, as they do for sparse, distant, or fast-moving objects, every method built on them fails together, and adding parameters or simulated training data does not fix what the formulation itself gets wrong. This paper takes the opposite path. We present CorrelationFlow, a training-free geometric framework that reduces scene flow to two textbook operations: connected-component labeling and correlation maximization on bird's-eye-view occupancy images. Objects are isolated as spatio-temporal connected components, their motions recovered as correlation peaks, and the resulting velocities propagated to all member points. However, this dense correlation evaluates every candidate displacement of every cluster and requires a window of past sweeps; therefore, we develop a sparse counterpart that operates on a single sweep pair by matching lightweight occupancy descriptors at boundary key points. Because nothing is trained, nothing is inherited: on the multi-domain test set of the Argoverse 2 2026 Scene Flow Challenge, spanning five datasets with heterogeneous sensors and platforms, CorrelationFlow ranked second among unsupervised methods and degrades most gracefully at long range, where the shared assumptions of learned methods break down. Our results suggest that a substantial share of the scene flow problem is solvable by classical computer vision, and that progress may require questioning the formulation, not scaling it.
