---
license: apache-2.0
language:
  - en
arxiv: 2605.30344
library_name: peft
base_model: Qwen/Qwen2.5-VL-7B-Instruct
datasets:
  - PLAN-Lab/VisAnomBench
pipeline_tag: image-text-to-text
tags:
  - anomaly-detection
  - time-series
  - vision-language
  - explainability
  - lora
  - peft
---

# VisAnomReasoner-7B

**Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection**

**Xiaona Zhou, Muntasir Wahed, Tianjiao Yu, Constantin Brif, Ismini Lourentzou**

### 🎉 Accepted to NeurIPS 2026

[📄 Paper](https://arxiv.org/abs/2605.30344) · [🌐 Project Page](https://plan-lab.github.io/projects/VisAnom/) · [💻 Code](https://github.com/PLAN-Lab/VisAnom)

[🤗 VisAnomBench](https://huggingface.co/datasets/PLAN-Lab/VisAnomBench) · [🤗 VisAnomReasoner-3B](https://huggingface.co/PLAN-Lab/VisAnomReasoner-3B)

---

## Overview

**VisAnomReasoner-7B** is a parameter-efficient vision-language model for time-series anomaly detection. It is a LoRA adapter for Qwen2.5-VL-7B-Instruct, fine-tuned on **VisAnomBench** to jointly localize anomalies and explain them from time-series plots.

> **Given a time-series plot, VisAnomReasoner locates the anomalous intervals and explains the visual evidence behind each decision.**

<p align="center">
  <img src="VisAnomReasoner_teaser.png" width="100%">
</p>

Left: Given a time-series plot, VisAnomReasoner locates anomalies with details grounded in the plot. Right: VisAnomReasoner-7B outperforms the strongest baselines on VisAnomBench and TSB-AD-U.

---

## Key Idea

Public time-series anomaly detection benchmarks typically provide interval annotations but not natural-language rationales, which makes it difficult to fine-tune VLMs to produce grounded, interpretable decisions.

**VisAnomBench** fills this gap by augmenting four public benchmarks with temporally grounded explanations, selected from multiple large VLMs using fine-grained, task-specific rewards. Fine-tuning on this explanation-augmented supervision teaches a compact VLM to couple interval-level localization with step-by-step reasoning grounded in what is visible in the plot.

---

## Highlights

* **Explanation-augmented supervision:** Trained on VisAnomBench, which pairs anomaly intervals with step-by-step explanations for time series from KPI, GutenTAG, UCR-EGI, and UCR-TSAD.

* **Joint localization and explanation:** Outputs anomaly intervals in a parseable format together with a numbered reasoning trace.

* **Parameter-efficient:** A rank-32 LoRA adapter with 95.2M trainable parameters (1.13% of the backbone) on top of Qwen2.5-VL-7B-Instruct.

* **Strong localization:** On VisAnomBench, both VisAnomReasoner variants achieve the highest precision, recall, F1, and overlap against 15 baselines, including general-purpose VLMs, LLM/VLM anomaly detectors, time-series foundation models, and classical detectors. VisAnomReasoner-7B improves overlap by 11 percentage points over its base model, Qwen2.5-VL-7B.

* **Cross-benchmark generalization:** On TSB-AD-U, which contains time series with unseen characteristics and anomaly definitions, VisAnomReasoner-7B improves standard precision and F1 by 9.57 and 13.39 percentage points over the strongest baseline, Qwen2.5-VL-7B.

---

## Resources

* **Paper:** [Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection](https://arxiv.org/abs/2605.30344)
* **Project Page:** [plan-lab.github.io/projects/VisAnom](https://plan-lab.github.io/projects/VisAnom/)
* **Code:** [github.com/PLAN-Lab/VisAnom](https://github.com/PLAN-Lab/VisAnom)
* **Dataset:** [PLAN-Lab/VisAnomBench](https://huggingface.co/datasets/PLAN-Lab/VisAnomBench)

---

## Citation

If you find VisAnomReasoner useful in your research, please consider citing our work:

```bibtex
@inproceedings{zhou2026tiny,
  title={Tiny but Trusted: Efficient Vision-Language Reasoning for Time-Series Anomaly Detection},
  author={Zhou, Xiaona and Wahed, Muntasir and Yu, Tianjiao and Brif, Constantin and Lourentzou, Ismini},
  booktitle={Advances in Neural Information Processing Systems (NeurIPS)},
  year={2026}
}
```