ONER: Online Experience Replay for Incremental Anomaly Detection

TL;DR

ONER introduces an efficient online experience replay framework for incremental anomaly detection, effectively preventing catastrophic forgetting and feature conflicts, and achieving state-of-the-art results with minimal parameters and training time.

Contribution

The paper proposes ONER, a novel framework combining decomposed prompts and semantic prototypes to enhance incremental anomaly detection performance.

Findings

Achieves +4.4% Pixel AUROC improvement on MVTec AD

Achieves +28.3% Pixel AUPR improvement on MVTec AD

Uses only 0.019M parameters and 5 training epochs per task

Abstract

Incremental anomaly detection aims to sequentially identify defects in industrial product lines but suffers from catastrophic forgetting, primarily due to knowledge overwriting during parameter updates and feature conflicts between tasks. In this work, We propose ONER (ONline Experience Replay), an end-to-end framework that addresses these issues by synergistically integrating two types of experience: (1) decomposed prompts, which dynamically generate image-conditioned prompts from reusable modules to retain prior knowledge thus prevent knowledge overwriting, and (2) semantic prototypes, which enforce separability in latent feature spaces at pixel and image levels to mitigate cross-task feature conflicts. Extensive experiments demonstrate the superiority of ONER, achieving state-of-the-art performance with +4.4% Pixel AUROC and +28.3% Pixel AUPR improvements on the MVTec AD dataset over prior methods. Remarkably, ONER achieves this with only 0.019M parameters and 5 training epochs per task, confirming its efficiency and stability for real-world industrial deployment.