SpiritSight Agent: Advanced GUI Agent with One Look

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Huang, Zhiyuan, Cheng, Ziming, Pan, Junting, Hou, Zhaohui, Zhan, Mingjie
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912330321756160
author Huang, Zhiyuan
Cheng, Ziming
Pan, Junting
Hou, Zhaohui
Zhan, Mingjie
author_facet Huang, Zhiyuan
Cheng, Ziming
Pan, Junting
Hou, Zhaohui
Zhan, Mingjie
contents Graphical User Interface (GUI) agents show amazing abilities in assisting human-computer interaction, automating human user's navigation on digital devices. An ideal GUI agent is expected to achieve high accuracy, low latency, and compatibility for different GUI platforms. Recent vision-based approaches have shown promise by leveraging advanced Vision Language Models (VLMs). While they generally meet the requirements of compatibility and low latency, these vision-based GUI agents tend to have low accuracy due to their limitations in element grounding. To address this issue, we propose $\textbf{SpiritSight}$, a vision-based, end-to-end GUI agent that excels in GUI navigation tasks across various GUI platforms. First, we create a multi-level, large-scale, high-quality GUI dataset called $\textbf{GUI-Lasagne}$ using scalable methods, empowering SpiritSight with robust GUI understanding and grounding capabilities. Second, we introduce the $\textbf{Universal Block Parsing (UBP)}$ method to resolve the ambiguity problem in dynamic high-resolution of visual inputs, further enhancing SpiritSight's ability to ground GUI objects. Through these efforts, SpiritSight agent outperforms other advanced methods on diverse GUI benchmarks, demonstrating its superior capability and compatibility in GUI navigation tasks. Models and datasets are available at https://hzhiyuan.github.io/SpiritSight-Agent.
format Preprint
id arxiv_https___arxiv_org_abs_2503_03196
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SpiritSight Agent: Advanced GUI Agent with One Look
Huang, Zhiyuan
Cheng, Ziming
Pan, Junting
Hou, Zhaohui
Zhan, Mingjie
Computer Vision and Pattern Recognition
Human-Computer Interaction
Robotics
Graphical User Interface (GUI) agents show amazing abilities in assisting human-computer interaction, automating human user's navigation on digital devices. An ideal GUI agent is expected to achieve high accuracy, low latency, and compatibility for different GUI platforms. Recent vision-based approaches have shown promise by leveraging advanced Vision Language Models (VLMs). While they generally meet the requirements of compatibility and low latency, these vision-based GUI agents tend to have low accuracy due to their limitations in element grounding. To address this issue, we propose $\textbf{SpiritSight}$, a vision-based, end-to-end GUI agent that excels in GUI navigation tasks across various GUI platforms. First, we create a multi-level, large-scale, high-quality GUI dataset called $\textbf{GUI-Lasagne}$ using scalable methods, empowering SpiritSight with robust GUI understanding and grounding capabilities. Second, we introduce the $\textbf{Universal Block Parsing (UBP)}$ method to resolve the ambiguity problem in dynamic high-resolution of visual inputs, further enhancing SpiritSight's ability to ground GUI objects. Through these efforts, SpiritSight agent outperforms other advanced methods on diverse GUI benchmarks, demonstrating its superior capability and compatibility in GUI navigation tasks. Models and datasets are available at https://hzhiyuan.github.io/SpiritSight-Agent.
title SpiritSight Agent: Advanced GUI Agent with One Look
topic Computer Vision and Pattern Recognition
Human-Computer Interaction
Robotics
url https://arxiv.org/abs/2503.03196