arXiv cs.AI· Jiannan Huang, Jerry Ding, Li Yu, Zhenting Li·· 1 天前AI 评分31
WebOverlay：面向极端宽高比网页浮层检测的高吞吐纯视觉蒸馏方案
WebOverlay: High-Throughput Vision-Only Distillation for Extreme Aspect-Ratio Web Overlay Detection
AI 导读
WebOverlay 是一个纯视觉网页浮层检测器，通过从前沿视觉语言教师模型蒸馏训练，无需解析 DOM 树即可识别 cookie 同意横幅、订阅提示等遮挡元素。它采用动态切块策略将极端宽高比的整页渲染切成保持宽高比的重叠图块，并配合宽高比感知的边界判定机制消除拼接伪影。该方法显著优于基于 DOM 的启发式规则，精度与召回率可媲美大型专有多模态模型，同时吞吐量高出数个数量级、部署成本大幅降低。
正文 · 原文
Abstract:Web overlays, such as cookie consent banners, newsletter prompts, and age gates, obstruct automated web crawlers and degrade downstream document representations. While proprietary multimodal Large Language Models (MLLMs) can accurately identify overlays, their latency and inference cost preclude web-scale ingestion. Conversely, lightweight vision models struggle with full-page renderings characterized by extreme vertical aspect ratios, where standard isotropic resizing collapses critical text features and causes false positives. In this work, we present WebOverlay, a high-throughput vision-only detector trained via distillation from a frontier vision-language teacher. We establish a dynamic tiling strategy that partitions extreme aspect-ratio web renders into aspect-preserving patches with spatial overlap, coupled with an aspect-ratio-aware boundary adjudication mechanism to eliminate boundary artifacts. To eliminate cross-modal latency overheads, our student model relies purely on visual cues, bypassing DOM tree parsing. WebOverlay significantly outperforms DOM-based heuristics and achieves competitive precision and recall against heavy proprietary MLLM baselines while delivering multiple orders of magnitude higher throughput and substantially reduced deployment costs.
| Comments: | 9 pages, 2 figures, 2 tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR) |
| ACM classes: | I.4.8; H.3.5; I.2.10 |
| Cite as: | arXiv:2610.41037 [cs.CV] |
| (or arXiv:2610.41037v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41037 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jiannan Huang [view email]
[v1]
Thu, 29 Oct 2026 22:29:33 UTC (918 KB)
来源：arXiv cs.AI · arxiv.org