Learning Multi-view Anomaly Detection with Efficient Adaptive Selection

TL;DR

This paper introduces a novel Multi-View Anomaly Detection (MVAD) method with an adaptive selection algorithm that effectively integrates multi-view features, achieving state-of-the-art results with efficient computation on the Real-IAD dataset.

Contribution

The paper proposes a new MVAD approach with an adaptive feature selection algorithm that improves multi-view anomaly detection accuracy and efficiency.

Findings

Achieved +2.5 improvement over previous methods on 10 metrics

Validated effectiveness on Real-IAD dataset with state-of-the-art performance

Reduced computational complexity to O((hw)^4/3)

Abstract

This study explores the recently proposed and challenging multi-view Anomaly Detection (AD) task. Single-view tasks will encounter blind spots from other perspectives, resulting in inaccuracies in sample-level prediction. Therefore, we introduce the Multi-View Anomaly Detection (MVAD) approach, which learns and integrates features from multi-views. Specifically, we propose a Multi-View Adaptive Selection (MVAS) algorithm for feature learning and fusion across multiple views. The feature maps are divided into neighbourhood attention windows to calculate a semantic correlation matrix between single-view windows and all other views, which is an attention mechanism conducted for each single-view window and the top-k most correlated multi-view windows. Adjusting the window sizes and top-k can minimise the complexity to O((hw)^4/3). Extensive experiments on the Real-IAD dataset under the multi-class setting validate the effectiveness of our approach, achieving state-of-the-art performance with an average improvement of +2.5 across 10 metrics at the sample/image/pixel levels, using only 18M parameters and requiring fewer FLOPs and training time. The codes are available at https://github.com/lewandofskee/MVAD.