Abstract:Image anomaly detection (AD) is widely researched on many occasions in computer vision tasks. High-dimensional data, such as image data, with noise and complex background is still challenging to detect anomalies under the situation that imbalanced or incomplete data are available. Some deep learning methods can be trained in an unsupervised way and map the original input into low-dimensional manifolds to predict larger differences in anomalies according to normal ones by dimension reduction. However, training a single low-dimension latent space is limited to present the low-dimensional features due to the fact that the noise and irreverent features are mapped into this space, resulting in that the manifolds are not discriminative for detecting anomalies. To address this problem, a new autoencoder framework is proposed in this study with two trainable mutually orthogonal complementary subspaces in the latent space, by latent subspace projection (LSP) mechanism, which is named as LSP-CAE. Specifically, latent subspace projection is used to train the latent image subspace (LIS) and the latent kernel subspace (LKS) in the latent space of the autoencoder-like model respectively, which can enhance learning power of different features from the input instance. The features of normal data are projected into the latent image subspace, while the latent kernel subspace is trained to extract the irrelevant information according to normal features by end-to-end training. To verify the generality and effectiveness of the proposed method, we replace the convolutional network with the fully-connected network contucted in the real-world medical datasets. The anomaly score based on projection norms in two subspace is used to evaluate the anomalies in the testing. Consequently, our proposed method can achieve the best performance according to four public datasets in comparison of the state-of-the-art methods.

Anomaly Detection in Chest X-ray Images with Adversarial Masked Autoencoder

A Weakly-Supervised Anomaly Detection Method Via Adversarial Training for Medical Images

Auto-AD: Autonomous Hyperspectral Anomaly Detection Network Based on Fully Convolutional Autoencoder

AMAE: Adaptation of Pre-Trained Masked Autoencoder for Dual-Distribution Anomaly Detection in Chest X-Rays

Unsupervised Anomaly Detection in Medical Images with a Memory-augmented Multi-level Cross-attentional Masked Autoencoder

Unsupervised Deep Anomaly Detection for Medical Images Using an Improved Adversarial Autoencoder

Anomaly detection by using a combination of generative adversarial networks and convolutional autoencoders

Anomaly Detection for Medical Images using Heterogeneous Auto-Encoder

Abnormal ECG detection based on an adversarial autoencoder

Dual-Constraint Autoencoder and Adaptive Weighted Similarity Spatial Attention for Unsupervised Anomaly Detection

Masked Autoencoders for Unsupervised Anomaly Detection in Medical Images

Convolutional autoencoder based on latent subspace projection for anomaly detection

3D Masked Autoencoders with Application to Anomaly Detection in Non-Contrast Enhanced Breast MRI

Anomaly Detection in Chest X-rays Based on Dual-Attention Mechanism and Multi-Scale Feature Fusion

A Novel MAE-Based Self-Supervised Anomaly Detection and Localization Method

Hyperspectral Anomaly Detection Based on Spatial-Spectral Cross Guided Mask Auto-Encoder

Dual-Distribution Discrepancy for Anomaly Detection in Chest X-Rays

Abnormal Chest X-Ray Identification With Generative Adversarial One-Class Classifier

PatchCL-AE: Anomaly detection for medical images using patch-wise contrastive learning-based auto-encoder

Rethinking Autoencoders for Medical Anomaly Detection from A Theoretical Perspective

Position-Guided Prompt Learning for Anomaly Detection in Chest X-Rays