MoViAD: Modular Visual Anomaly Detection

2025-07-16

AI Summary

Visual Anomaly Detection (VAD) faces core challenges including severe scarcity of anomalous samples, heavy reliance on unsupervised learning paradigms, and constraints in edge deployment. To address these, we propose a unified, modular, open-source VAD framework supporting complex training scenarios—such as continual learning, few-shot adaptation, and noise-robust training—while natively enabling efficient deployment on resource-constrained edge and IoT devices. The framework decouples model architectures, backbone networks, evaluation metrics, and deployment tooling, and integrates pixel-level/image-level evaluation, model quantization, compression, and comprehensive performance profiling. It further enables seamless customization of models and datasets. To our knowledge, this is the first work to standardize and modularize the entire VAD pipeline—from training and evaluation to lightweight deployment—thereby significantly lowering both algorithmic development and engineering deployment barriers. Empirical results demonstrate improved inference efficiency and practical utility on edge hardware.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Learning on the Edge & Model Compression
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

VAD is a critical field in machine learning focused on identifying deviations from normal patterns in images, often challenged by the scarcity of anomalous data and the need for unsupervised training. To accelerate research and deployment in this domain, we introduce MoViAD, a comprehensive and highly modular library designed to provide fast and easy access to state-of-the-art VAD models, trainers, datasets, and VAD utilities. MoViAD supports a wide array of scenarios, including continual, semi-supervised, few-shots, noisy, and many more. In addition, it addresses practical deployment challenges through dedicated Edge and IoT settings, offering optimized models and backbones, along with quantization and compression utilities for efficient on-device execution and distributed inference. MoViAD integrates a selection of backbones, robust evaluation VAD metrics (pixel-level and image-level) and useful profiling tools for efficiency analysis. The library is designed for fast, effortless deployment, enabling machine learning engineers to easily use it for their specific setup with custom models, datasets, and backbones. At the same time, it offers the flexibility and extensibility researchers need to develop and experiment with new methods.

Problem

• Addresses scarcity of anomalous data in visual anomaly detection
• Provides modular tools for diverse VAD scenarios and deployments
• Enables flexible research and efficient practical implementation

Innovation

• Modular library for visual anomaly detection
• Supports Edge and IoT with optimized models
• Integrates diverse backbones and evaluation metrics