O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning
2.80T1 sourcearXiv cs.MA
arXiv cs.MAoriginal source ↗
Source record
Published by arXiv cs.MA (T1 source). The original is at https://arxiv.org/abs/2607.18142.
Pipeline notes
The summary and note below are generated by the signal pipeline — they are Beyond Desk’s reading, not quotations from the source.
SummaryIntroduces O-VAD, a training-free agentic framework for industrial video anomaly detection that tracks spatial-temporal dynamics of objects and reasons over object-wise temporal state trajectories to identify anomalies, outperforming frontier VLMs and existing methods on three IVAD datasets.
Why it mattersTraining-free design with object-state trajectory reasoning is a transferable pattern for agentic systems, though the industrial video focus narrows direct workstation relevance.
Cited by
No citations on record.