用少量参数实现高效弱监督视频异常检测，精度媲美大模型。
ProDisc-VAD: An Efficient System for Weakly-Supervised Anomaly Detection in Video Surveillance Applications
- 通过可学习原型控制正常样本建模，避免被多数正常数据淹没。
- 仅对最可信的极端得分样本进行对比学习，提升异常与正常区分度。
- 参数量仅0.4M，适合部署在资源受限的监控场景中。
弱监督视频异常检测（WS-VAD）常因标签模糊导致特征学习困难。本文提出ProDisc-VAD框架，包含两个协同组件：原型交互层（PIL）利用少量可学习原型实现受控的正常性建模，建立稳健基线；伪实例判别增强损失（PIDE）对得分最高/最低的可靠极端实例实施针对性对比学习，提升类别可分性。该方法在ShanghaiTech和UCF-Crime数据集上分别取得97.98%和87.12%的AUC，参数量仅为0.4M，远低于基于ViT的方法（如VadCLIP，超过800倍更少）。代码已开源。
原文摘要 · Abstract (English)
Weakly-supervised video anomaly detection (WS-VAD) using Multiple Instance Learning (MIL) suffers from label ambiguity, hindering discriminative feature learning. We propose ProDisc-VAD, an efficient framework tackling this via two synergistic components. The Prototype Interaction Layer (PIL) provides controlled normality modeling using a small set of learnable prototypes, establishing a robust baseline without being overwhelmed by dominant normal data. The Pseudo-Instance Discriminative Enhancement (PIDE) loss boosts separability by applying targeted contrastive learning exclusively to the most reliable extreme-scoring instances (highest/lowest scores). ProDisc-VAD achieves strong AUCs (97.98% ShanghaiTech, 87.12% UCF-Crime) using only 0.4M parameters, over 800x fewer than recent ViT-based methods like VadCLIP. Code is available at https://github.com/modadundun/ProDisc-VAD.
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。