---
格式版本: 2
标题: "Test-Time Self-Evolving GUI Visual Grounding via Reflection-Guided On-Policy Self-Distillation"
原文链接: "https://arxiv.org/abs/2608.11191"
发布日期: "2026-08-11"
发布时间校准状态: "found"
发布时间需复核: "否"
发布时间来源: "rule:local:strict_original_body"
发布时间证据: "**\\[v1\\]** Tue, 11 Aug 2026 17:50:25 UTC (534 KB)"
发布时间校准原因: "规则确认唯一严格发布时间，来源 local:strict_original_body"
发布时间校准置信度: "high"
发布时间候选数量: 18
发布时间严格候选数量: 6
发布时间原页读取状态: "source template page reused from URL open"
发布时间未找到原因: ""
发布时间校准时间: "2026-08-12T11:15:11+08:00"
发布时间仲裁状态: "skipped"
发布时间仲裁尝试次数: 0
发布时间仲裁耗时毫秒: 0
发现时间: "2026-08-12T11:13:54+08:00"
入库时间: "2026-08-12T03:15:11.252Z"
来源平台: "arXiv 学术论文搜索"
搜索渠道: "source_template"
搜索词: "https://arxiv.org/search/?query=deployment&searchtype=all"
匹配关键词:
  - "deployment"
  - "AI"
相关厂家:
  []
相关专家:
  []
内容类型: "网页"
抓取工具: "Free Fetch + Defuddle"
清洗工具: "Defuddle Markdown + Defuddle/Readability 正文提取"
原始附件:
  []
AI优质: "否"
AI打分: 15
AI分档: "非优质"
AI质检状态: "不通过"
AI打分理由: "论文内容为GUI视觉定位的测试时自适应方法，完全不涉及超节点、AI Rack、机柜级AI基础设施或相关硬件、供电、散热、互连等主题，与项目关注范围无关。"
AI质检模型: "ali-deepseek-v4-flash"
AI质检时间: "2026-08-12T11:15:22+08:00"
AI主题相关性: 0
AI来源权威性: 5
AI新颖性: 0
AI技术细节: 0
AI商业部署信号: 0
AI完整性: 10
采集批次: "2026年8月12日2点04分20秒"
采集批次ID: "20260812-020420-289"
去重键: "https://arxiv.org/abs/2608.11191"
---

## Computer Science > Computer Vision and Pattern Recognition

## Title:Test-Time Self-Evolving GUI Visual Grounding via Reflection-Guided On-Policy Self-Distillation

Authors:[Shiyu Xuan](https://arxiv.org/search/cs?searchtype=author&query=Xuan,+S), [Zechao Li](https://arxiv.org/search/cs?searchtype=author&query=Li,+Z)

[View PDF](https://arxiv.org/pdf/2608.11191) [HTML (experimental)](https://arxiv.org/html/2608.11191v1)

> Abstract:GUI Visual Grounding is a fundamental capability for GUI agents. Existing models typically freeze their parameters after deployment, limiting their ability to adapt to unseen interfaces. Although recent methods attempt to adapt models via test-time reinforcement learning, they cannot reflect upon failed exploration. To overcome this, we propose a Test-Time Self-Evolving framework that enables models to improve after deployment without human-annotated ground truth. It constructs a closed-loop of Exploration, Evaluation, Reflection, and Internalization. Specifically, the agent first explores unseen interfaces by predicting grounding coordinates for given instructions. To evaluate these explorations, we introduce an MLLM-based Reflector to assess the generated results and provide the corresponding reasoning reflections. To internalize reflection knowledge into the model weights, we propose Reflection-Guided On-Policy Self-Distillation, which translates high-level reasoning into dense token-level supervision via a conditioned self-teacher. Furthermore, we design a Contrastive Calibration method to prevent incorrect auto-regressive prefixes from corrupting the supervisory signals during failed explorations. Extensive experiments across six benchmarks demonstrate our framework's effectiveness, achieving an average accuracy improvement of 7.4% over the base model. To the best of our knowledge, this is the first work to successfully exploit on-policy self-distillation for test-time adaptation in GUI visual grounding. By filling the gap in post-deployment adaptation, our framework completes the self-evolving capability of GUI agents. The code will be released.

| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| --- | --- |
| Cite as: | [arXiv:2608.11191](https://arxiv.org/abs/2608.11191) \[cs.CV\] |
|  | (or [arXiv:2608.11191v1](https://arxiv.org/abs/2608.11191v1) \[cs.CV\] for this version) |
|  | [https://doi.org/10.48550/arXiv.2608.11191](https://doi.org/10.48550/arXiv.2608.11191) |

## Submission history

From: Shiyu Xuan \[[view email](https://arxiv.org/show-email/c06a9eed/2608.11191)\]  
**\[v1\]** Tue, 11 Aug 2026 17:50:25 UTC (534 KB)

[Which authors of this paper are endorsers?](https://arxiv.org/auth/show-endorsers/2608.11191) | Disable MathJax ([What is MathJax?](https://info.arxiv.org/help/mathjax.html))