Sequential PatchCore: Anomaly Detection for Surface Inspection using Synthetic Impurities

TL;DR

This paper introduces Sequential PatchCore, a scalable anomaly detection method for surface inspection that leverages synthetic data with impurities, improving detection performance on high-resolution images.

Contribution

It presents a novel sequential coreset training approach enabling anomaly detection on large images and emphasizes the importance of synthetic impurities in data generation.

Findings

Synthetic data with impurities improves model training.

Sequential PatchCore enables training on high-res images with consumer hardware.

Finetuning on real data enhances detection performance.

Abstract

The appearance of surface impurities (e.g., water stains, fingerprints, stickers) is an often-mentioned issue that causes degradation of automated visual inspection systems. At the same time, synthetic data generation techniques for visual surface inspection have focused primarily on generating perfect examples and defects, disregarding impurities. This study highlights the importance of considering impurities when generating synthetic data. We introduce a procedural method to include photorealistic water stains in synthetic data. The synthetic datasets are generated to correspond to real datasets and are further used to train an anomaly detection model and investigate the influence of water stains. The high-resolution images used for surface inspection lead to memory bottlenecks during anomaly detection training. To address this, we introduce Sequential PatchCore - a method to build coresets sequentially and make training on large images using consumer-grade hardware tractable. This allows us to perform transfer learning using coresets pre-trained on different dataset versions. Our results show the benefits of using synthetic data for pre-training an explicit coreset anomaly model and the extended performance benefits of finetuning the coreset using real data. We observed how the impurities and labelling ambiguity lower the model performance and have additionally reported the defect-wise recall to provide an industrially relevant perspective on model performance.