https://github.com/greyhatguy007/hazeclear
https://github.com/greyhatguy007/hazeclear
Last synced: about 1 year ago
JSON representation
- Host: GitHub
- URL: https://github.com/greyhatguy007/hazeclear
- Owner: greyhatguy007
- Created: 2023-09-20T16:46:30.000Z (almost 3 years ago)
- Default Branch: main
- Last Pushed: 2023-09-20T17:49:05.000Z (almost 3 years ago)
- Last Synced: 2025-05-09T00:09:14.082Z (over 1 year ago)
- Language: Jupyter Notebook
- Size: 12.7 KB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# AI-ML based Intelligent de-hazing algorithm
## Introduction
This project aims to develop a deep learning model to dehaze images. Image dehazing is pivotal for enhancing visual clarity in various applications in computer vision tasks. The model is constructed using an encoder-decoder architecture in PyTorch.
## Dataset
We are utilizing the [Dense-Haze dataset](https://www.kaggle.com/datasets/rajat95gupta/hazing-images-dataset-cvpr-2019) available on Kaggle. This dataset contains 55 pairs of hazy and non-hazy images, making it a suitable benchmark for image dehazing tasks.
## Project Steps
1. **Data Preprocessing**:
- Resize images and normalize pixel values.
2. **Model Development**:
- Design an encoder-decoder neural network architecture.
- Initialize model layers, including convolutional layers, pooling layers, and upsampling layers.
- Define the forward propagation mechanism.
3. **Training the Model**:
- Define loss function (Mean Squared Error) and optimizer (Adam).
- Train the model using the training dataset.
4. **Model Evaluation**:
- Compute evaluation metrics, including Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index (SSIM).
5. **Results Visualization**:
- Showcase original, hazy, and dehazed images side by side.
- Plot training loss and validation loss across epochs.
6. **Optimization (if necessary)**:
- Tune hyperparameters, such as learning rate, batch size, and number of epochs.
- Modify the model architecture by adding more layers or changing layer configurations.
- Implement data augmentation to artificially increase the training dataset size and robustness.
7. **Deployment**:
- Convert the trained model into a format suitable for deployment.
- Integrate the model into a web or mobile application for real-time image dehazing.
## Technical Details
- **Framework**: PyTorch
- **Model Architecture**: [Encoder-Decoder CNN](./docs/encoder-decoder-architecture/readme.md)
- **Loss Function**: Mean Squared Error (MSE)
- **Optimizer**: Adam
- **Evaluation Metrics**: [PSNR, SSIM](./docs/evaluation-metric/readme.md)
## Future Work
1. Experiment with advanced architectures like U-Net or ResNet.
2. Implement a custom loss function for improved performance.
3. Enhance the deployment solution to process video streams for real-time dehazing.
## Contributions
Contributions to this project are welcome. Please ensure that you adhere to the project's coding and documentation standards.
## Acknowledgments
Special thanks to the creators of the Dense-Haze dataset for making it publicly available on Kaggle.