Colorization using Optimization

Python and C++ implementations of a user-guided image/video colorization technique as proposed by the paper Colorization Using Optimization. The algorithm is based on a simple premise; neighboring pixels in space-time that have similar intensities should have similar colors. This premise is formalized using a quadratic cost function that obtains an optimization problem that can be solved efficiently using standard techniques. While using this alogorithm, an artist only needs to annotate the image with a few color scribbles or visual clues, and the indicated colors are automatically propagated in both space and time to produce a fully colorized image or sequence. The annotation can be done using any drawing tool such as JSPaint or Gimp.

Instructions

Instructions for running python version

1. Create a virtualenv using:

   • virtualenv venv --python=python3
   • source venv/bin/activate
   • pip install -r requirements.txt

2. Colorize images using the CLI:

   python colorize.py
   
   Options:
       --original_image TEXT    Original Image Path
       --visual_clue TEXT       Visual Clue Image Path
       --result_path TEXT       Colorized Image Path (without file extensions)
       -i, --use_itercative     Use Iterative Mode
       --epochs INTEGER         Number of epochs for Iterative Mode
       --log_intervals INTEGER  Log Interval
       --help                   Show this message and exit.
   

3. Alternatively, you can run on Google Colab using

Instructions to build C++ version

1. Install dependencies using sh install.sh

2. Create a build directory mkdir build && cd build

3. Generate makefiles and compile using cmake .. && make

4. Run the executable using ./colorization [input-image] [visual-clues] [result] [gamma] [threshold]

5. Alternatively, you can download the executable from here and run it (installation of dependencies is still needed).

Results

Original Image	Visual Clues	Colorized Image