Awesome

Image and video colorizer

Image and video colorizer is package for automatic image and video colorization. Models are already trained.

Instalation

Installation can be done in 5 easy steps

Install all requirements for Tensorflow and tensorflow itself with:
```
pip install tensorflow-gpu
```
if you use GPU device for computation otherwise:
```
pip install tensorflow
```

Create virtual environment

virtualenv -p python3 colorization_venv

Activate virtual environment
```
source colorization_venv/bin/activate
```

Clone Image and video colorization package and move in it

git clone https://github.com/PrimozGodec/ImageColorization.git
cd ImageColorization

Install requirements
```
pip install -r requirements.txt
```
You are done :)

In case you do not have a GPU device in your computer, please install Tensorflow for a CPU. Instructions are at the Tnesorflow website.

Image colorization

For automatic image colorizing follow those steps:

Copy images into /data/image/original directory
Run main.py script from src/image_colorization/ directory.
```
python -m src.image_colorization.main --model <model name>
```
Parameter --method is optional, if not present reg_full_model is default. It can be choose from this list:
- reg_full_model (default)
- reg_full_vgg_model
- reg_part_model
- class_weights_model
- class_wo_weights_model
You can find colored images in /data/image/colorized directory.

on your GPU or CPU specifications. You will see progress bar that show you how far you are with colorization.

Video colorization

For automatic video colorizing follow those steps:

Copy images into /data/video/original directory
Run video_colorizer.py script from src/video_colorization/ directory.
```
python -m src.video_colorization.video_colorizer
```
Video colorizer is always using reg_full_model.
You can find colored videos in /data/video/colorized directory.

Colorization take few hours since there is a lot of images to color in a video and depends on your GPU or CPU specifications and length of a video. You will see progress bar that show you how far you are with colorization.