Official announcements from Microsoft or updates on Hugging Face

Microsoft Launches OmniParser Model for UI Automation on Hugging Face, MIT License, High Accuracy, Outperforming GPT-4 Vision

Oct 25, 2024, 03:39 AM

Microsoft has introduced a new AI model named OmniParser, designed to enhance the capabilities of UI agents by converting UI screenshots into a structured format. Released on the Hugging Face platform, OmniParser is described as a general screen parsing tool that improves existing large language model (LLM) based UI agents. The model aims to address current limitations in screen parsing techniques and boasts high accuracy in benchmarking tests. It is noted for its ability to handle various document formats and is licensed under the MIT license, making it accessible for web automation applications. Notably, OmniParser reportedly outperforms the GPT-4 Vision model in screen understanding benchmarks, indicating its potential effectiveness in the field of AI automation.