PEDENet: Image Anomaly Localization via Patch Embedding and Density Estimation
This addresses anomaly detection in images for applications like quality control, but appears incremental as it builds on existing techniques like Gaussian Mixture Models.
The paper tackles unsupervised image anomaly localization by proposing PEDENet, which uses patch embedding, density estimation, and location prediction networks, achieving performance benchmarked against state-of-the-art methods without specifying concrete numbers.
A neural network targeting at unsupervised image anomaly localization, called the PEDENet, is proposed in this work. PEDENet contains a patch embedding (PE) network, a density estimation (DE) network, and an auxiliary network called the location prediction (LP) network. The PE network takes local image patches as input and performs dimension reduction to get low-dimensional patch embeddings via a deep encoder structure. Being inspired by the Gaussian Mixture Model (GMM), the DE network takes those patch embeddings and then predicts the cluster membership of an embedded patch. The sum of membership probabilities is used as a loss term to guide the learning process. The LP network is a Multi-layer Perception (MLP), which takes embeddings from two neighboring patches as input and predicts their relative location. The performance of the proposed PEDENet is evaluated extensively and benchmarked with that of state-of-the-art methods.