用可变形注意力与因果建模，实现实时高精度视频异常检测
VigilFormer: Deformable Attention for Video Anomaly Detection with Causal Risk Inference
- 采用稀疏可变形时空注意力，降低计算开销同时捕捉不规则运动
- 在三个数据集上分别达到87.83%、97.21%、89.74%的AUC，推理速度达41.5 FPS
- 适合部署于实时监控场景，尤其适用于带宽或算力受限环境
监控场景下的视频异常检测需在准确率与实时吞吐间取得平衡。现有方法通常通过增强特征提取器或优化架构来解决，但很少兼顾两者。本文提出VigilFormer，融合可变形时空注意力与因果时间建模，实现对未剪辑监控视频的异常检测。其提出的稀疏可变形时空编码器（DSTE）在帧间关注少数关键位置，避免密集注意力的二次方复杂度，仍能捕捉不规则运动模式。因果异常分类器（CAC）使用扩张因果卷积处理片段级特征，并通过对比多实例学习目标分离异常与正常表征，无需帧级标签。为满足部署需求，自适应置信度调度器（ACS）在推理时动态跳过低信息帧，减少静态场景中的冗余计算。在UCF-Crime、ShanghaiTech和CUHK Avenue上，VigilFormer分别取得87.83%、97.21%、89.74%的AUC，在单张GPU上达到41.5 FPS，优于近期弱监督方法的准确率与速度表现。
原文摘要 · Abstract (English)
Video anomaly detection in surveillance settings must balance detection accuracy against real-time throughput, a tension that existing methods address either through stronger feature extractors or more efficient architectures, but rarely both. We present VigilFormer, a unified framework that combines deformable spatio-temporal attention with causal temporal modeling to detect anomalies in untrimmed surveillance video. The proposed Deformable Spatio-Temporal Encoder (DSTE) attends to a sparse set of informative locations across frames, avoiding the quadratic cost of dense attention while retaining the ability to capture irregular motion patterns. A Causal Anomaly Classifier (CAC) applies dilated causal convolutions over snippet-level features and optimizes a contrastive multiple-instance learning objective that separates anomalous and normal representations without frame-level labels. To meet deployment constraints, an Adaptive Confidence Scheduler (ACS) dynamically skips low-information frames at inference time, reducing redundant computation in static scenes. Evaluated on UCF-Crime, ShanghaiTech, and CUHK Avenue, VigilFormer achieves AUC scores of 87.83%, 97.21%, and 89.74% respectively, at 41.5 FPS on a single GPU, outperforming recent weakly-supervised methods in both accuracy and speed.
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。