PT to TFJS Model Converter

This repository contains a Python application that uses YOLO to detect license plates and exports the model to various formats including ONNX and TensorFlow.js.

Setup Instructions

Prerequisites

• Python 3.8 or higher
• pip (Python package installer)

Installation

1. Clone this repository or download the source code.

2. Create and activate a virtual environment (recommended):

   python -m venv venv
   
   # On macOS/Linux
   source venv/bin/activate
   
   # On Windows
   venv\Scripts\activate

3. Install the required dependencies:

   pip install -r requirements.txt

Running the Application

To run the application, simply execute the app.py script:

python app.py

This will:

1. Load the YOLO model (your-model.pt)
2. Export the model to ONNX format
3. Attempt to export to TensorFlow SavedModel format
4. Convert the model to TensorFlow.js format

Output Files

After running the application, you should expect the following output files:

• your-model.onnx: The ONNX version of the model
• your-model_saved_model/: Directory containing the TensorFlow SavedModel
• tfjs_model/: Directory containing the TensorFlow.js model

Troubleshooting

If you encounter any issues:

1. Make sure all dependencies are correctly installed
2. Check that the model file your-model.pt exists in the project directory
3. Ensure you have sufficient disk space for the exported models

License

This project is licensed under the ISC License.