This is a novel methodology to perform dehazing process on a single outdoor Image using feature extraction techniques in Deep Learning and 3 added pre-processing steps.
Dataset used for training and validation is SOTS outdoor available here.
Detailed explanation and documentation available here
Follow the following steps to run the code:
- Download the image datasets from kaggle- link :- https://www.kaggle.com/datasets/wwwwwee/dehaze
- Create separate folders for each specific dataset type: hazy images, clear images and hazy test images.
- Open the VGG16 CNN model on Google colab/Google colab pro
- Run the code (provide proper dataset path for input to the model)
- Please note that you may have to run the VGG16 model network function code-block again once the model is trained (for executing pruning without error).
- Provide the path for hazy test images to the trained model and run the block of code.
- Dehazed image will be generated by the model.