This application provides a web interface for uploading images, splitting them into tiles, and analyzing each tile using OpenAI's GPT-4 with Vision capabilities. Users can specify the size of the tiles and provide a prompt for the analysis. The application also supports searching for specific terms within the analysis results.
From my blog post: How to use GPT-4 with Vision for Robotics and Other Applications
- Image Upload: Users can upload images either by dragging and dropping or by using the file input.
- Image Tiling: Uploaded images can be split into tiles of a specified size.
- Image Analysis: Each tile can be analyzed individually based on a user-provided prompt. The analysis leverages OpenAI's GPT-4 with Vision.
- Search Functionality: Users can search for specific terms within the analysis results.
- Ensure Python and Flask are installed.
- Install required Python packages:
flask,requests,flask_cors,python-dotenv,werkzeug. - Set up an .env file with your OpenAI API key as OPENAI_API_KEY.
- Start the Flask server by running
python app.py. - Access the web interface at
http://localhost:5007.
- Uploading an Image: Use the "Image" input field to select an image file.
- Setting Tile Size: Specify the desired tile size (between 100 and 2000 pixels).
- Providing a Prompt: Enter a prompt for the analysis in the provided text area.
- Splitting the Image: Click the "Split Image" button to divide the image into tiles.
- Analyzing Tiles: Use the "Analyze" button to analyze all tiles.
- Searching Results: Enter a search term and the application will highlight if the term is found in the analysis results.