Skip to content

HastingsGreer/ImageInpainting

Repository files navigation

Requirements:

numpy keras tensorflow PIL opencv

demo.py will download the pretrained weights, then continuously process "target.png" in the project directory, by masking out and then inpainting all pixels of the color (0, 255, 0), and displaying to a window.

Workflow:

Open an image in your preferred editor.

Save as target.png in the project directory

Execute demo.py with python3

Draw on image in bright green (0, 255, 0).

Save image to run inpainting algorithm

Docker Workflow (CPU only):

Run demo.py once to download network weights.

Build dockerfile with command sudo docker build -t inpainting .

Process an image with command

sudo docker run -v ~/full_path_to_image_dir/:/img/ inpainting /img/target.png /img/out.png

Releases

No releases published

Packages

No packages published