| # Debluring Application |
|
|
| This application provides tools for deblurring images using deep learning. The system consists of: |
|
|
| 1. A **DeblurGAN module** for image deblurring using deep learning |
| 2. A **FastAPI backend** for processing images via API endpoints |
| 3. A **Streamlit frontend** for a user-friendly web interface |
|
|
| ## Features |
|
|
| - Upload and deblur images with a single click |
| - Modern web interface with side-by-side comparison of original and deblurred images |
| - Download deblurred results |
| - GPU-accelerated processing (with fallback to CPU) |
| - Command-line interface for batch processing |
|
|
| ## Requirements |
|
|
| - Python 3.7+ |
| - PyTorch |
| - CUDA (optional, for GPU acceleration) |
| - Other dependencies listed in requirements.txt |
|
|
| ## Setup Instructions |
|
|
| 1. Clone this repository: |
|
|
| ``` |
| git clone <repo-url> |
| cd Debluring |
| ``` |
|
|
| 2. Install dependencies: |
|
|
| ``` |
| pip install -r requirements.txt |
| ``` |
|
|
| **Note**: This application requires the `pretrainedmodels` package for the deblurring model. If you encounter any issues, you can manually install the required packages: |
|
|
| ``` |
| pip install pretrainedmodels torchsummary albumentations opencv-python-headless |
| ``` |
|
|
| ## Running the Application |
|
|
| You can run the application in three different ways: |
|
|
| ### 1. Standalone DeblurGAN Module |
|
|
| Process all images in the `inputs/` directory and save results to `outputs/`: |
|
|
| ``` |
| python deblur_module.py |
| ``` |
|
|
| ### 2. API Server |
|
|
| Start the FastAPI server: |
|
|
| ``` |
| python api.py |
| ``` |
|
|
| The API will be available at: http://localhost:8001 |
|
|
| - POST `/deblur/` - Upload an image for deblurring |
| - GET `/status/` - Check API status |
| - GET `/diagnostics/` - Get system diagnostics |
|
|
| ### 3. Web Interface |
|
|
| First, ensure the API server is running, then start the Streamlit interface: |
|
|
| ``` |
| python app.py |
| ``` |
|
|
| The web UI will be available at: http://localhost:8501 |
|
|
| ### 4. All-in-One (Optional) |
|
|
| Use the run.py script to start both the API server and the Streamlit interface: |
|
|
| ``` |
| python run.py |
| ``` |
|
|
| ## Using the Application |
|
|
| 1. **Command-line mode**: |
|
|
| - Place your blurry images in the `inputs/` directory |
| - Run `python deblur_module.py` |
| - Find the deblurred results in the `outputs/` directory |
|
|
| 2. **Web interface mode**: |
| - Start both the API and the Streamlit app |
| - Upload an image through the web interface |
| - Click "Deblur Image" and wait for processing |
| - Download the deblurred image when ready |
|
|
| ## Troubleshooting |
|
|
| If you encounter any issues: |
|
|
| 1. Check the API diagnostics endpoint: http://localhost:8001/diagnostics/ |
|
|
| 2. Review logs in the `logs/` directory for detailed error information. |
|
|
| 3. Make sure your system has sufficient resources for running the deblurring model. |
|
|
| 4. If you're experiencing CUDA errors, the application will automatically fall back to CPU processing. |
|
|
| 5. For large images, the application will automatically resize them to a maximum dimension of 2000 pixels to avoid memory issues. |
|
|
| ## License |
|
|
| This project is licensed under the MIT License - see the LICENSE file for details. |
|
|
| ## Acknowledgements |
|
|
| This application is based on the DeblurGAN architecture for image deblurring with modifications to create a user-friendly interface and robust error handling. |
|
|
