AuraTracer智迹闻
中文

EVENT DOSSIER

Adaptive Multi-Granularity Temporal Modeling for Weakly Supervised Video Anomaly Detection

2026-09-07 12:00 Models 🔥 40.2 heat score
1sources
1days unfolding
40.2heat score
1mentions
SummaryAI generated

On September 7, 2026, arXiv cs.AI published a research paper titled “Adaptive Multi-Granularity Temporal Modeling for Weakly Supervised Video Anomaly Detection”. This study proposes an adaptive multi-granularity temporal modeling approach aimed at addressing the challenges in weakly supervised video anomaly detection. By integrating multi-granularity information with temporal dynamics, this method improves the model’s ability to identify anomalies and its generalization performance in the absence of dense labeled data.

Related eventsRELATED EVENTS
Key entitiesKEY ENTITIES
arXiv

SignalsSIGNALS

Keyword heat
  • arXiv1

All reports (1)SOURCES

A arXiv cs.AI en 2026-09-07 12:00

Adaptive Multi-Granularity Temporal Modeling for Weakly Supervised Video Anomaly Detection

提出了一种用于弱监督视频异常检测(WSVAD)的自适应多粒度时间建模框架。该框架包含三个核心模块:利用动态位置编码和可学习类别 token 构建稳定全局表示的时间细化模块(TRM)、通过时间不连续性分析识别事件边界并聚合特征的自适应事件分割模块(ESM),以及将异常分数动态整合到视频级预测的基于相似度的融合策略。实验表明,该方法在两个基准测试中表现优于最先进方法,有效解决了现有方法因依赖刚性手工时间先验而导致对异常时长和动态变化适应不足的问题。