Single Dehazing Image using Deep Learning
This project presented a encoder-decoder type neural net- work, that directly removes haze given a single image of the scene without explicitly using the atmospheric scattering model. We employed the usage of the GMAN and a parallel network to improve the process of single image dehazing. We provided a quantitative and visual comparison of our model, proving the superior quality of our model. We had evaluated our model on over 500 hazed and dehazed image sets and compared to the models in the market, our model faired better. Our model works for both indoor and outdoor images.
To Run ===>
- pip install -r requiremts.txt
- cd src
- python my_example.py
- cd ..
- streamlit run app.py