Saved in:
Bibliographic Details
Main Authors: Gu, Yao, Xu, Xiaohao, Wu, Yingna
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.15237
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918391344791552
author Gu, Yao
Xu, Xiaohao
Wu, Yingna
author_facet Gu, Yao
Xu, Xiaohao
Wu, Yingna
contents Vision-Language Models (VLMs) demonstrate strong general-purpose reasoning but remain limited in physics-grounded anomaly detection, where causal understanding of dynamics is essential. Existing VLMs, trained predominantly on appearance-centric correlations, fail to capture kinematic constraints, leading to poor performance on anomalies such as irregular rotations or violated mechanical motions. We introduce a physics-informed instruction tuning framework that explicitly encodes object properties, motion paradigms, and dynamic constraints into structured prompts. By delivering these physical priors through multi-turn dialogues, our method decomposes causal reasoning into incremental steps, enabling robust internal representations of normal and abnormal dynamics. Evaluated on the Phys-AD benchmark, our approach achieves 96.7% AUROC in video-level detection--substantially outperforming prior SOTA (66.9%)--and yields superior causal explanations (0.777 LLM score). This work highlights how structured physics priors can transform VLMs into reliable detectors of dynamic anomalies.
format Preprint
id arxiv_https___arxiv_org_abs_2603_15237
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Multi-turn Physics-informed Vision-language Model for Physics-grounded Anomaly Detection
Gu, Yao
Xu, Xiaohao
Wu, Yingna
Computer Vision and Pattern Recognition
Vision-Language Models (VLMs) demonstrate strong general-purpose reasoning but remain limited in physics-grounded anomaly detection, where causal understanding of dynamics is essential. Existing VLMs, trained predominantly on appearance-centric correlations, fail to capture kinematic constraints, leading to poor performance on anomalies such as irregular rotations or violated mechanical motions. We introduce a physics-informed instruction tuning framework that explicitly encodes object properties, motion paradigms, and dynamic constraints into structured prompts. By delivering these physical priors through multi-turn dialogues, our method decomposes causal reasoning into incremental steps, enabling robust internal representations of normal and abnormal dynamics. Evaluated on the Phys-AD benchmark, our approach achieves 96.7% AUROC in video-level detection--substantially outperforming prior SOTA (66.9%)--and yields superior causal explanations (0.777 LLM score). This work highlights how structured physics priors can transform VLMs into reliable detectors of dynamic anomalies.
title Multi-turn Physics-informed Vision-language Model for Physics-grounded Anomaly Detection
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.15237