license: mit
This dataset is a filtered version of the omniact dataset, containing only grounding tasks. It is intended as an example dataset for the training of our InfiGUI-G1 model. For more details on our work, please see our GitHub repository: https://github.com/InfiXAI/InfiGUI-G1.
Dataset Details
The omniact_grounding_filtered dataset is derived from the original omniact dataset, with the following modifications:
- Grounding Tasks Only: We have selected only the grounding task samples.
- Hard Samples Only: We filtered out overly simple samples. Specifically, any sample that was solved correctly in all 8 attempts (i.e., had a 100% success rate) was removed from the dataset.
Citation Information
If you find this dataset or our work useful in your research, we would be grateful if you consider citing our work.
@article{liu2025infiguig1,
title={InfiGUI-G1: Advancing GUI Grounding with Adaptive Exploration Policy Optimization},
author={Liu, Yuhang and Liu, Zeyu and Zhu, Shuanghe and Li, Pengxiang and Xie, Congkai and Wang, Jiasheng and Hu, Xueyu and Han, Xiaotian and Yuan, Jianbo and Wang, Xinyao and others},
journal={arXiv preprint arXiv:2508.05731},
year={2025}
}
We also strongly recommend citing the original data source, as their work was foundational to ours:
@misc{kapoor2024omniact,
title={OmniACT: A Dataset and Benchmark for Enabling Multimodal Generalist Autonomous Agents for Desktop and Web},
author={Raghav Kapoor and Yash Parag Butala and Melisa Russak and Jing Yu Koh and Kiran Kamble and Waseem Alshikh and Ruslan Salakhutdinov},
year={2024},
eprint={2402.17553},
archivePrefix={arXiv},
primaryClass={cs.AI}
}
Acknowledgements
We extend our sincere gratitude to the authors and contributors of the original omniact dataset for their significant work and for making their valuable data publicly available. Their efforts have been foundational for research in this area.