COFT-AD: COntrastive Fine-Tuning for Few-Shot Anomaly Detection

arXiv:2402.18998v1 Announce Type: new
Abstract: Existing approaches towards anomaly detection~(AD) often rely on a substantial amount of anomaly-free data to train representation and density models. However, large anomaly-free datasets may not always be available before the inference stage; in which case an anomaly detection model must be trained with only a handful of normal samples, a.k.a. few-shot anomaly detection (FSAD). In this paper, we propose a novel methodology to address the challenge of FSAD which incorporates two important techniques. Firstly, …

abstract anomaly anomaly detection arxiv case cs.cv data datasets detection few-shot fine-tuning free free data free datasets inference representation stage train type