Couleurs

Couleurs is a simple image colorizing project. The application is supposed to receive inputs in form of images, then return the colorized same image. This project utilizes the model of DeOldify by Jason Antic as the AI to colorize images.

Team Members

• Rachell Vannessa Christian as Project Coordinator
• Kalista Gabriela Willies as Technical and Research Analyst
• Monica Agustina Chandra as UI and UX Designer
• Michael Onasis Hasri as Major Front-end Developer
• Wesley Aldrich as Back-end Developer, Minor Front-end Developer

Website

In this repository, I'm creating the web application using Flask. The goal of this project is:

• To make sure I completely understand how DeOldify AI works.
• To simply learn how to use Flask framework in building simple web applications.

How to use

Clone this repository into your device, then run the "app.py" file from your terminal. If any error pops up, it might be because some required libraries are not in your device. Install all the dependencies that are "unrecognized".

Cloning this repository

git clone https://github.com/wesleyaldrich/Couleurs.git
cd Couleurs

"Couleurs" is the web application root directory, which is now cloned into your device.

Installing DeOldify Model

• Download an AI Model Weights for our image colorizer.
• In this project, we will use the Artistic Completed Generator Weights from DeOldify.
  Download Here
• After it is downloaded, the file should be named ColorizeArtistic_gen.pth. Replace the existing dummy file in Couleurs/ai/models/!
• Thus, the file path for the model weights should be Couleurs/ai/models/ColorizeArtistic_gen.pth.

Installing dependencies

If your Windows device is compatible with Python 3.7.6, you can install the required libraries perfectly by running:

pip install -r requirements_freeze.txt

Otherwise, you can run this:

pip install -r requirements.txt

Or manually by:

pip install flask matplotlib pandas scipy requests fastprogress torch torchvision opencv-python ffmpeg yt_dlp ipython

If you're using Linux/Ubuntu, you might also need:

sudo apt install libgl1-mesa-glx

• Note: Consider using virtual environment as well.

Run the web application

• For Linux/Ubuntu:

  python3 app.py
  

• For Windows:

  python app.py
  

• Or you can adjust with your own operating system or preference.

By default, you should see Couleurs local web app running in http://127.0.0.1:8080/.

---

• DeOldify by Jason Antic https://github.com/jantic/DeOldify/