Online-Adaptive Anomaly Detection for Defect Identification in Aircraft Assembly

TL;DR

This paper introduces an online-adaptive anomaly detection framework for aircraft defect identification that leverages transfer learning and feature similarity measures to improve detection accuracy in various environments.

Contribution

It presents a novel transfer learning-based framework that adapts to different environments by selecting similar training images and fitting a normality model for anomaly detection.

Findings

Detection accuracy exceeds 0.975

Outperforms the state-of-the-art ET-NET approach

Effective across multiple benchmarks and laboratory data

Abstract

Anomaly detection deals with detecting deviations from established patterns within data. It has various applications like autonomous driving, predictive maintenance, and medical diagnosis. To improve anomaly detection accuracy, transfer learning can be applied to large, pre-trained models and adapt them to the specific application context. In this paper, we propose a novel framework for online-adaptive anomaly detection using transfer learning. The approach adapts to different environments by selecting visually similar training images and online fitting a normality model to EfficientNet features extracted from the training subset. Anomaly detection is then performed by computing the Mahalanobis distance between the normality model and the test image features. Different similarity measures (SIFT/FLANN, Cosine) and normality models (MVG, OCSVM) are employed and compared with each other. We evaluate the approach on different anomaly detection benchmarks and data collected in controlled laboratory settings. Experimental results showcase a detection accuracy exceeding 0.975, outperforming the state-of-the-art ET-NET approach.