Anomaly detection through latent space restoration using vector-quantized variational autoencoders
This work provides an incremental improvement for unsupervised anomaly detection in images, particularly for the MOOD challenge datasets.
This paper proposes an out-of-distribution detection method using Vector-Quantized Variational Auto-Encoders (VQ-VAEs) combined with an Auto-Regressive (AR) model to estimate prior probabilities of latent codes. The method defines sample-wise anomaly scores based on negative log-likelihood and pixel-wise scores using L1 distance between original and restored images, achieving higher accuracies on MOOD challenge datasets compared to standard VAE reconstruction.
We propose an out-of-distribution detection method that combines density and restoration-based approaches using Vector-Quantized Variational Auto-Encoders (VQ-VAEs). The VQ-VAE model learns to encode images in a categorical latent space. The prior distribution of latent codes is then modelled using an Auto-Regressive (AR) model. We found that the prior probability estimated by the AR model can be useful for unsupervised anomaly detection and enables the estimation of both sample and pixel-wise anomaly scores. The sample-wise score is defined as the negative log-likelihood of the latent variables above a threshold selecting highly unlikely codes. Additionally, out-of-distribution images are restored into in-distribution images by replacing unlikely latent codes with samples from the prior model and decoding to pixel space. The average L1 distance between generated restorations and original image is used as pixel-wise anomaly score. We tested our approach on the MOOD challenge datasets, and report higher accuracies compared to a standard reconstruction-based approach with VAEs.