基准AUC不能代表部署可靠性：面向监控视频异常检测的现成特征的跨数据集审计
Benchmark AUC Is Not Deployable Reliability: A Cross-Dataset Audit of Off-the-Shelf Features for Surveillance Video Anomaly Detection
AI总结：
本研究通过跨数据集实验发现，监控视频异常检测的基准AUC在跨场景部署时降至随机水平，表明现有方法不可靠。
摘要：
自动“可疑行为”标记是AI监控的一个标志性承诺，该领域在标准视频异常检测基准上报告了高帧级ROC-AUC。这些数字是通过在同一摄像头和场景上训练和测试得到的。我们审计了当放弃这一假设时会发生什么。我们使用一个数据集的所有正常训练帧，采用冻结的现成嵌入（CLIP、DINOv2、ResNet-50、EfficientNet-B0）和最近邻距离，构建无监督正常性模型，并对同一数据集和其他数据集的测试帧进行评分。在4个真实数据集（UCSD Ped1、UCSD Ped2、CUHK Avenue、ShanghaiTech）和4个骨干网络上，同数据集AUC平均为0.704，但跨数据集AUC平均为0.499，即随机水平：在一个场景上校准的检测器在另一个场景上不比抛硬币好，并且在几个数据对中低于随机水平。最强的骨干网络使情况更糟，而非更好：DINOv2具有最佳的同数据集AUC（在Ped2上高达0.901）和最大的跨数据集下降。这种崩溃并非评分规则的伪影：用PaDiM风格的马氏距离检测器替换最近邻检测器几乎完全重现了结果（跨数据集差距0.202 vs 0.208）。即使在有利的工作点，误报率也高达每小时31,931次。我们得出结论，监控异常检测引用的基准数字描述的是校准的实验室环境，大大夸大了部署可靠性，并发布了重现所有数字的代码。
英文摘要：
Automated "suspicious behavior" flagging is a headline promise of AI surveillance, and the field reports high frame-level ROC-AUC on standard video anomaly detection benchmarks. Those numbers are measured by training and testing on the same camera and scene. We audit what happens when that assumption is dropped. We build an unsupervised normality model from the all-normal training frames of one dataset, using frozen off-the-shelf embeddings (CLIP, DINOv2, ResNet-50, EfficientNet-B0) and a nearest-neighbour distance, and score the test frames of the same and of other datasets. Across 4 real datasets (UCSD Ped1, UCSD Ped2, CUHK Avenue, ShanghaiTech) and 4 backbones, same-dataset AUC averages 0.704 but cross-dataset AUC averages 0.499, which is chance: a detector calibrated on one scene is no better than a coin flip on another, and in several pairs it is below chance. The strongest backbone makes this worse, not better: DINOv2 has the best same-dataset AUC (up to 0.901 on Ped2) and the largest cross-dataset drop. The collapse is not an artefact of the scoring rule: replacing the nearest-neighbour detector with a PaDiM-style Mahalanobis detector reproduces it almost exactly (cross-dataset gap 0.202 versus 0.208). Even at a favourable operating point the false-alarm rate is on the order of 31,931 per hour. We conclude that the benchmark numbers quoted for surveillance anomaly detection describe a calibrated laboratory setting and overstate deployable reliability by a wide margin, and we release the code that reproduces every number.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
