Lens Flare Reduction by Flare Simulation and U-Net
- Prepare Image Set (Image Set Used in Training: https://www.kaggle.com/arnaud58/landscape-pictures)
- Apply Flare Simulation to Image Set using flare_simulation.py
- Create Dataloader using dataloader.py
- Run train.py to start training
- Run predict.py to predict flare area and inpaint flared image
- .ipynb file is for Colab environment
- Final model available in models
- Training structure inspired by: https://towardsdatascience.com/creating-and-training-a-u-net-model-with-pytorch-for-2d-3d-semantic-segmentation-dataset-fb1f7f80fe55