Testing out various ML projects Blur and Clear is based off of (a fork and edited copy of) https://github.com/aditya9211/Blur-and-Clear-Classification

The jupyter notebook has the most detailed comments - explains the quick start

*The original scripts have not been changed! if you just run those you will probably not get the same results

• yml was added - this is for creating a conda env, will still have to pip install the requirements.txt
• import functions rather than running the scripts - move through the notebook this way and can change the config.py hyper-parameters in the notebook
• args have been removed for the train.py and given hard encoding in the notebook
• in the predict.py part there is a new ability to direct the model to a folder of images - based on prediction the script will copy the image to a corresponding folder. *can change the shutil.copy to shutil.move to save disk space if you are sure about the results.
• some %%time commands were inserted so you can see how long each training of the model and prediction takes