MVP: Multiple View Prediction Improves GUI Grounding

TL;DR

This paper introduces Multi-View Prediction (MVP), a training-free framework that improves GUI grounding accuracy by aggregating multiple cropped views, significantly enhancing stability and performance across various models and benchmarks.

Contribution

MVP is a novel, training-free approach that leverages multi-view inference with attention-guided view proposals and clustering to stabilize coordinate predictions in GUI grounding.

Findings

MVP improves accuracy across multiple models and benchmarks.

MVP significantly boosts UI-TARS-1.5-7B to 56.1%.

MVP enhances stability against visual perturbations.

Abstract

GUI grounding, which translates natural language instructions into precise pixel coordinates, is essential for developing practical GUI agents. However, we observe that existing grounding models exhibit significant coordinate prediction instability, minor visual perturbations (e.g. cropping a few pixels) can drastically alter predictions, flipping results between correct and incorrect. This instability severely undermines model performance, especially for samples with high-resolution and small UI elements. To address this issue, we propose Multi-View Prediction (MVP), a training-free framework that enhances grounding performance through multi-view inference. Our key insight is that while single-view predictions may be unstable, aggregating predictions from multiple carefully cropped views can effectively distinguish correct coordinates from outliers. MVP comprises two components: (1) Attention-Guided View Proposal, which derives diverse views guided by instruction-to-image attention scores, and (2) Multi-Coordinates Clustering, which ensembles predictions by selecting the centroid of the densest spatial cluster. Extensive experiments demonstrate MVP's effectiveness across various models and benchmarks. Notably, on ScreenSpot-Pro, MVP boosts UI-TARS-1.5-7B to 56.1%, GTA1-7B to 61.7%, Qwen3VL-8B-Instruct to 65.3%, and Qwen3VL-32B-Instruct to 74.0%. The code is available at https://github.com/ZJUSCL/MVP.