计算机工程与科学2026，Vol.48Issue(8)：1433-1444,12.DOI:10.3969/j.issn.1007-130X.2026.08.010
基于条件扩散概率模型的视频异常检测
A conditional diffusion model for video anomaly detection
摘要
Abstract
Video anomaly detection is becoming increasingly important in modern society.Because there are various fuzzy behaviors in videos and the categories cannot be exhaustive,it is difficult to de-fine normal and abnormal based on single-classification methods.In response to the aforementioned is-sues,a conditional diffusion probabilistic model for video anomaly detection(CDiffuVAD)is proposed.Firstly,a normal content generator is designed to improve the content accuracy of the image generated by the model.It highlights the distribution pattern of normal samples through a memory pool and lever-ages the diffusion probability model to learn the complex distribution of video data.Secondly,implicit motion conditions are designed and introduced to learn the spatiotemporal features of video segments.The bidirectional optical flow information is used as the implicit motion condition of the diffusion process,and the coordinate normalization method is used to provide the coordinate embedding of the segment frame,so as to realize the learning and fitting of the motion trend in the multi-frame sequence data and alleviate the sensitivity of the model to the hard normal information in the video.Experimental results show that the proposed method achieves frame-level AUC scores of 85.7%on the Avenue data-set,75.5%on the ShanghaiTech dataset,and 65.7%on the UBnormal dataset,demonstrating its abili-ty to capture diverse features in normal samples and its effectiveness in video anomaly detection tasks.关键词
视频异常检测/扩散模型/不确定性/记忆池/生成模型Key words
video anomaly detection/diffusion model/uncertainty/memory pool/generative model分类
信息技术与安全科学引用本文复制引用
叶亚琴,汤子健,牛嘉诚,张新欢..基于条件扩散概率模型的视频异常检测[J].计算机工程与科学,2026,48(8):1433-1444,12.基金项目
湖北省智能地学信息处理重点实验室开放研究项目(KLIGIP-2022-B14) （KLIGIP-2022-B14）
桐兴(湖北)投资咨询有限公司项目(2021116490) （湖北）
