DerainNet– U-Net Autoencoder for Rain Removal
A deep learning-based image restoration system using U-Net that removes rain from images to enhance visual clarity in real-world applications.
📝 Notes
This project was developed as part of a research-driven initiative to explore image restoration using deep learning. It showcases how deep neural networks can effectively enhance image clarity in adverse weather conditions like rain.
📄 Project Description
This project addresses the challenge of removing rain streaks from images using a deep learning-based approach. A synthetic dataset was created using OpenCV by overlaying rain effects on clean images. The model leverages a U-Net architecture, built on an encoder-decoder framework with skip connections, to learn the mapping between rainy and clean images. Trained using TensorFlow on Google Colab, the model delivers accurate and visually appealing derained outputs. This solution is valuable for autonomous systems, surveillance, and photography where visual clarity is crucial.
🛠️ Languages & Tools Used
- Python
- TensorFlow / Keras
- OpenCV
- Google Colab
- NumPy
- Matplotlib
📊 Dataset
📦 Download:
Download Dataset👨💻 Project Members
- Nandana J – Model Design & Training
- Sandra Mariya George – Dataset Generation & Preprocessing
- Stiya Johnson – Evaluation & Visualization
- Vishnuhari V A- Model Optimization & Testing