Zoom in, Click out: Unlocking and Evaluating the Potential of Zooming for GUI Grounding

TL;DR

This paper introduces ZoomClick, a training-free method leveraging zoom properties to enhance GUI grounding, achieving state-of-the-art results and proposing a new benchmark for zoom adaptability in GUI models.

Contribution

It explores zoom as a prior for GUI grounding, proposes a novel zoom-based method, and introduces a benchmark to evaluate zoom adaptability in GUI grounding models.

Findings

ZoomClick significantly improves GUI grounding performance.

Achieved state-of-the-art results on multiple benchmarks.

Introduced GUIZoom-Bench for zoom adaptability evaluation.

Abstract

Grounding is a fundamental capability for building graphical user interface (GUI) agents. Although existing approaches rely on large-scale bounding box supervision, they still face various challenges, such as cross-platform generalization, complex layout analysis, and fine-grained element localization. In this paper, we investigate zoom as a strong yet underexplored prior for GUI grounding, and propose a training-free method, ZoomClick. By characterizing four key properties of zoom (i.e., pre-zoom, depth, shrink size, minimal crop size), we unlock its full capabilities for dynamic spatial focusing and adaptive context switching. Experiments demonstrate that our method significantly boosts the performance of both general vision-language and specialized GUI grounding models, achieving state-of-the-art results on several mainstream benchmarks; for example, UI-Venus-72B attains a 73.1% success rate on ScreenSpot-Pro. Furthermore, we present GUIZoom-Bench, a benchmark for evaluating model adaptability to zoom, aiming to inspire future research on improving zoom for further training and test-time scaling in GUI grounding tasks.