Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection
Summary
This paper introduces VisAnomReasoner, a parameter-efficient vision-language model fine-tuned on a novel benchmark (VisAnomBench) with natural-language rationales, achieving over 21pp improvement in precision and F1 for time-series anomaly detection and strong cross-benchmark generalization.
View Cached Full Text
Cached at: 05/29/26, 07:03 PM
Paper page - Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection
Source: https://huggingface.co/papers/2605.30344
Abstract
A parameter-efficient vision-language model is developed for time-series anomaly detection using a novel benchmark with natural-language rationales, achieving superior performance and generalization across multiple datasets.
Recent advances inVision-Language Models(VLMs) have achieved impressive performance across many tasks, yet prior studies report unsatisfactory performance when applying large language or multimodal models to finding abnormal patterns in sequential data. Publicanomaly detectionbenchmarks typically provide interval annotations but notnatural-language rationales, making it difficult to fine-tune VLMs to produce grounded, interpretable decisions. To address this gap, we construct VisAnomBench, a curated benchmark built from publictime-series datasetsand augmented with high-quality anomaly explanations selected from multiple large VLMs using fine-grained, task-specific rewards. Through fine-tuning on this benchmark, we develop VisAnomReasoner, a parameter-efficient VLM for time-seriesanomaly detection. Experimental results on VisAnomBench show that VisAnomReasoner achieves more accurate anomaly localization and consistently outperforms all baselines, with improvements of at least 21.23 and 23.87 percentage points in precision and F1, respectively. Additional experiments on the TSB-AD-U benchmark demonstrate strongcross-benchmark generalization, with VisAnomReasoner improving precision and F1 by 9.57 and 13.39 percentage points, respectively.
View arXiv pageView PDFProject pageAdd to collection
Models citing this paper0
No model linking this paper
Cite arxiv.org/abs/2605.30344 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.30344 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.30344 in a Space README.md to link it from this page.
Collections including this paper0
No Collection including this paper
Add this paper to acollectionto link it from this page.
Similar Articles
Glance, Scrutinize, and Think: Advancing Video Anomaly Detection from Training-Free to Agentic Reasoning
This paper presents a unified global-to-local paradigm for video anomaly detection, introducing a training-free framework (GtS) and a tool-augmented agentic reasoning method with reinforcement learning, along with a new benchmark VAGU-T and metric JeAUG.
VLMs are Good Teachers for Video Reasoning via Adaptive Test-Time Optimization
This paper introduces a paradigm where Vision-Language Models (VLMs) act as test-time teachers to guide Video Generation Models (VGMs) via differentiable rewards and LoRA optimization, achieving a 16.7-point average improvement on video reasoning benchmarks.
More Thought, Less Accuracy? On the Dual Nature of Reasoning in Vision-Language Models
This paper uncovers that prolonged reasoning in vision-language models can impair perceptual grounding, causing recognition failures on basic visual questions. It proposes Vision-Anchored Policy Optimization (VAPO) to steer reasoning toward visually grounded trajectories, achieving state-of-the-art performance with the VAPO-Thinker-7B model.
ChronoVision: Temporal Reasoning via Latent State Reconstruction
ChronoVision is a multimodal framework that improves temporal reasoning in vision-language models by aligning visual logic with latent imagery, using a Reconstructive Visual Head, ROI Attention Locating module, and reinforcement learning. It introduces the Vbvr-VQA dataset and achieves SOTA accuracy on temporal tracking benchmarks.
O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning
O-VAD introduces a training-free agentic framework for industrial video anomaly detection that tracks object state evolution over time and reasons over temporal trajectories to identify abnormal objects, outperforming existing VLM and VAD methods on three datasets.