IPAD: Industrial Process Anomaly Detection Dataset

arXiv:2404.15033v1 Announce Type: new
Abstract: Video anomaly detection (VAD) is a challenging task aiming to recognize anomalies in video frames, and existing large-scale VAD researches primarily focus on road traffic and human activity scenes. In industrial scenes, there are often a variety of unpredictable anomalies, and the VAD method can play a significant role in these scenarios. However, there is a lack of applicable datasets and methods specifically tailored for industrial production scenarios due to concerns regarding privacy and security. …

abstract anomaly anomaly detection arxiv cs.cv dataset detection focus human industrial ipad process role scale traffic type video