Beer-Lambert Guided Representation Learning for Unsupervised Anomaly Detection in Sub-THz Food Inspection Images

2026-06-15

AI Summary

This work addresses the limitations of existing unsupervised anomaly detection methods, which rely on RGB-pretrained representations and fail to capture the transmission characteristics inherent in sub-terahertz (Sub-THz) imaging, leading to suboptimal performance in food foreign-body inspection. To overcome this, the study introduces the Beer–Lambert law into unsupervised representation learning for the first time, proposing an attenuation decomposition module that leverages physics-guided attenuation reconstruction as a regularizer to refine student representations. The method integrates Sub-THz transmission imaging with a one-class anomaly detection framework and employs a Leave-One-Food-Out evaluation protocol to rigorously assess cross-category generalization. Experiments on the Inline-Food-Inspection-THz dataset demonstrate substantial improvements over current baselines, validating the approach’s effectiveness and generalization capability in real-world food inspection scenarios.

Abstract

Food manufacturing requires reliable inspection systems to detect foreign material contamination and maintain product safety. Sub-THz transmission imaging provides material-dependent attenuation characteristics that are useful for detecting low-density contaminants in food products. However, existing unsupervised anomaly detection methods mainly rely on RGB-pretrained visual representations, which may not adequately capture the transmission behavior of Sub-THz images. This paper proposes a Beer-Lambert guided representation learning framework for unsupervised anomaly detection in Sub-THz food inspection images. The proposed method introduces an attenuation decomposition module as an auxiliary regularization module that constrains student representations through attenuation reconstruction during training. In addition to the conventional one-class setting, we introduce a Leave-One-Food-Out protocol to evaluate generalization capability under unseen food categories. Experimental results on the Inline-Food-Inspection-THz dataset show that the proposed method improves overall anomaly detection performance over the baseline method.

Problem

• Sub-THz imaging
• unsupervised anomaly detection
• food inspection
• foreign material contamination
• attenuation characteristics

Innovation

• Beer-Lambert law
• Sub-THz imaging
• unsupervised anomaly detection
• attenuation decomposition
• representation learning