Readme not clear for Window

[PierrunoYT]

The readme is not clear enough how to install on Windows. I needed to ask a LLM to guide me. Also there is nothing to delete so I dont know why there are remove commands.

[HoustonMuzamhindo]

I agree. Going back and forth and also checking HuggingFace

[abughalib]

This is what I did to get it running on my AMD Windows Machine.

Installation on Windows

Pre-requisites

0. Install Python 3.12 if not installed.
1. Download Anaconda Distribution from Anaconda's official website or from winget winget install --Id Anaconda.Anaconda3.
2. Add Anaconda to your user Environment Variable if not added, It is usually located in C:\Users\$Env:UserName\anaconda3\condabin.

Installation Setup

1. Clone the Project & Change to the Project Directory.

   git clone https://github.com/microsoft/OmniParser; cd OmniParser
   

2. Create a new conda environment and install the dependencies.

   conda create -n "omni" python==3.12
   conda activate omni
   pip install -r requirements.txt
   

3. Clone the Vision model from Huggingface.

   git clone https://huggingface.co/microsoft/OmniParser-v2.0 weights
   

4. Modify the path to florence2 model in gradio_demo.py file to huggingface model path.

   You can use this one microsoft/Florence-2-base or other one from here.

5. Run the demo.

   python .\gradio_demo.py
   

Note: Make sure you have cuda installed or it would use CPU by default.

[yadong-lu]

@PierrunoYT @abughalib Thanks for the suggestion. Edited the Readme. Let me know if there is anything unclear