Normality Addition via Normality Detection in Industrial Image Anomaly Detection Models

TL;DR

This paper introduces a novel method called NAND for adjusting image anomaly detection models to recognize new normal patterns post-training, using normality detection with vision-language models to improve adaptability in industrial scenarios.

Contribution

The paper proposes a new 'normality addition' scenario and a method leveraging vision-language models to update anomaly detection models without retraining from scratch.

Findings

NAND effectively incorporates new normalities into existing models.

The method improves anomaly detection accuracy in the presence of changing normality criteria.

Evaluation on MVTec AD demonstrates the method's practical effectiveness.

Abstract

The task of image anomaly detection (IAD) aims to identify deviations from normality in image data. These anomalies are patterns that deviate significantly from what the IAD model has learned from the data during training. However, in real-world scenarios, the criteria for what constitutes normality often change, necessitating the reclassification of previously anomalous instances as normal. To address this challenge, we propose a new scenario termed "normality addition," involving the post-training adjustment of decision boundaries to incorporate new normalities. To address this challenge, we propose a method called Normality Addition via Normality Detection (NAND), leveraging a vision-language model. NAND performs normality detection which detect patterns related to the intended normality within images based on textual descriptions. We then modify the results of a pre-trained IAD model to implement this normality addition. Using the benchmark dataset in IAD, MVTec AD, we establish an evaluation protocol for the normality addition task and empirically demonstrate the effectiveness of the NAND method.