An Application of Image Denoising

🌺Complete Code of DnCNN, please go to https://github.com/cszn/DnCNN.

Environment

• Model Scope: Ubuntu 22.04 operating system
• CUDA: 12.1.0
• PyTorch: 2.3.1
• TensorFlow: 2.16.1

Details

1. Technique: Autoencoder and DnCNN
2. Training Dataset: MNIST (30,000 images)
3. Validation Dataset: (Skipped)
4. Testing Dataset: MNIST (10 images)
5. Initial Learning Rate: 0.001
6. Training Time: 10 minutes / 5 epochs
7. Batch Size: 64
8. Denoised Performance Metrics: PSNR and SSIM

Note: There are still some areas that could be improved.

---

This is a simple attempt. You can run the programs in TrainingCode/AutoDnCNN_remake.
If you encounter any issues, feel free to reach out to me anytime at cye79698@gmail.com.