Fast style transfer in tensorflow and some restoration techniques which uses pillow and OpenCV
NOTE : This works only with Python2
Step 1 : Create a new folder and cd into that.
Step 2 : Create a new virtual environment by running virtualenv -p python2 <your_env_name>
Step 3 : Activate the new environment by running source <your_env_name>/bin/activate
Step 4 : Clone this repository into that environment
Step 5 : run `pip install -r requirements.txt'
To try the (basic) image restoration, cd into the image_restore folder
Only three modes are available (denoise, unsharp and inpaint)
Run python restore.py --input <path_to_the_input_image> --output <path_to_save_the_output_image> --mode
if image inpainting has to be performed then an additional --mask <path_to_the_mask_image has to be provided for the mask image
.
To try the image style transfer, cd into the style_transfer folder
run python evaluate.py --checkpoint <path_to_the_saved_models> --in-path <path_to_input_image> --out-path <path to output image>
