MMVIAD: Multi-view Multi-task Video Understanding for Industrial Anomaly Detection

TL;DR

This paper introduces MMVIAD, a comprehensive multi-view video dataset for industrial anomaly detection, along with a benchmark and a novel two-stage model that significantly improves detection performance.

Contribution

The authors present the first continuous multi-view video dataset for industrial anomaly detection and develop a two-stage post-training pipeline that enhances model accuracy.

Findings

Current video MLLMs perform below human level on MMVIAD.

VISTA model improves average task score from 45.0 to 57.5 on MMVIAD-Unseen.

Source code is publicly available at the provided GitHub link.

Abstract

Industrial anomaly detection is critical for manufacturing quality control, yet existing datasets mainly focus on static images or sparse views, which do not fully reflect continuous inspection processes in real industrial scenarios. We introduce MMVIAD (Multi-view Multi-task Video Industrial Anomaly Detection), to the best of our knowledge the first continuous multi-view video dataset for industrial anomaly detection and understanding, together with a benchmark for multi-task evaluation. MMVIAD contains object-centric 2-second inspection clips with approximately 120 degrees of camera motion, covering 48 object categories, 14 environments, and 6 structural anomaly types. It supports anomaly detection, defect classification, object classification, and anomaly visible-time localization. Systematic evaluations on MMVIAD show that current commercial and open-source video MLLMs remain far below human performance, especially for fine-grained defect recognition and temporal grounding. To improve transferable anomaly understanding, we further develop a two-stage post-training pipeline where PS-SFT (Perception-Structured Supervised Fine-Tuning) initializes perception-structured reasoning and VISTA-GRPO (Visibility-grounded Industrial Structured Temporal Anomaly Group Relative Policy Optimization) refines the model with semantic-gated defect reward and visibility-aware temporal reward, producing the final model VISTA. On MMVIAD-Unseen, VISTA improves the base model's average score across the four tasks from 45.0 to 57.5, surpassing GPT-5.4. Source code is available at https://github.com/Georgekeepmoving/MMVIAD.