Visual Anomaly Detection for Images: A Survey
Jie Yang, Ruijie Xu, Zhiquan Qi, Yong Shi University of Chinese Academy of Sciences, Beijing 101408, China, Email:
Abstract
Visual anomaly detection is an important and challenging problem in the field of machine learning and computer vision. This problem has attracted a considerable amount of attention in relevant research communities. Especially in recent years, the development of deep learning has sparked an increasing interest in the visual anomaly detection problem and brought a great variety of novel methods. In this paper, we provide a comprehensive survey of the classical and deep learning-based approaches for visual anomaly detection in the literature. We group the relevant approaches in view of their underlying principles and discuss their assumptions, advantages, and disadvantages carefully. We aim to help the researchers to understand the common principles of visual anomaly detection approaches and identify promising research directions in this field.
中文速览
无监督视觉异常检测（visual anomaly detection）长期面临一个核心难题：异常样本极为稀缺且形态多变，无法依赖有标注数据来训练检测模型，只能从正常数据中学习"什么是正常"，再去判断未知样本是否偏离正常。这篇综述系统梳理了从传统机器学习到深度学习时代的各类方法，将其归纳为密度估计、单类分类、图像重建和自监督分类四大类，并逐一分析每类方法的原理、优势与局限。研究覆盖图像级和像素级两种检测粒度，既回顾了高斯模型、支持向量数据描述等经典方法，也详细介绍了自编码器、生成对抗网络、流模型等深度学习方案的最新进展。这项工作的价值在于为研究者提供了一张清晰的"路线图"，帮助他们快速把握该领域的发展脉络，并在工业缺陷检测、医学影像病灶识别、智能安防等实际应用场景中找到合适的技术路径与未来突破口。
原文 arXiv:2109.13157;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2109.13157v1