Deep Anomaly Detection Using Geometric Transformations

TL;DR

This paper introduces a novel deep learning approach for image anomaly detection by training a model to recognize geometric transformations, enabling effective out-of-distribution image identification with improved accuracy.

Contribution

The paper proposes a new method that trains a neural network to identify geometric transformations, which enhances anomaly detection capabilities beyond existing techniques.

Findings

Outperforms state-of-the-art anomaly detection methods

Effective in detecting out-of-distribution images

Uses geometric transformations as auxiliary tasks

Abstract

We consider the problem of anomaly detection in images, and present a new detection technique. Given a sample of images, all known to belong to a "normal" class (e.g., dogs), we show how to train a deep neural model that can detect out-of-distribution images (i.e., non-dog objects). The main idea behind our scheme is to train a multi-class model to discriminate between dozens of geometric transformations applied on all the given images. The auxiliary expertise learned by the model generates feature detectors that effectively identify, at test time, anomalous images based on the softmax activation statistics of the model when applied on transformed images. We present extensive experiments using the proposed detector, which indicate that our algorithm improves state-of-the-art methods by a wide margin.