Incremental dimension reduction for efficient and accurate visual anomaly detection

TL;DR

This paper introduces an incremental dimension reduction method that efficiently reduces high-dimensional features for visual anomaly detection, enabling faster training without significant accuracy loss.

Contribution

The paper presents a batch-based incremental SVD algorithm that reduces feature dimensionality efficiently for large-scale visual anomaly detection tasks.

Findings

Accelerates training of anomaly detection models

Maintains high accuracy with reduced features

Handles large datasets efficiently

Abstract

While nowadays visual anomaly detection algorithms use deep neural networks to extract salient features from images, the high dimensionality of extracted features makes it difficult to apply those algorithms to large data with 1000s of images. To address this issue, we present an incremental dimension reduction algorithm to reduce the extracted features. While our algorithm essentially computes truncated singular value decomposition of these features, other than processing all vectors at once, our algorithm groups the vectors into batches. At each batch, our algorithm updates the truncated singular values and vectors that represent all visited vectors, and reduces each batch by its own singular values and vectors so they can be stored in the memory with low overhead. After processing all batches, we re-transform these batch-wise singular vectors to the space spanned by the singular vectors of all features. We show that our algorithm can accelerate the training of state-of-the-art anomaly detection algorithm with close accuracy.