Paper | Project | Presentation
This repository contains source codes which used for "Final Project for CUAI 3rd Conference".
CUAI 3rd Final Conference 1st place Award 🏆 (Grand Prize)
Paper and Presentation are in Korean.
- Byunghyun Bae (School of Pharmaceutics, Chung-Ang Univ.)
- Hearyeon Seo (School of Mechanical Engineering, Chung-Ang Univ.)
- Nahyuk Lee 🙋♂️ (School of Computer Science & Engineering, Chung-Ang Univ.)
- Bo-lim Lee (School of Computer Science & Engineering, Chung-Ang Univ.)
- Hayun Lee (School of Computer Science & Engineering, Chung-Ang Univ.)
- Whanjin Lee (School of Energy Systems Engineering, Chung-Ang Univ.)
We use 2 emotion class 'happy' and 'sad' for test, and our System can generate paintings depend on user emotions.
The system flow of our project is as follows.
Measuring brain waves data and Emotional classification with them. Recommending reference paintings according to the emotional evaluation results.
Rendering paintings for target image using SinGAN.
We recommend you to use Anaconda that already including mandotory packages.
python -m pip install -r requirements.txt
Our code was tested with Python 3.6, Pytorch 1.7, CUDA 11.
To train emotional classification models with your own brain waves dataset('ratio.csv'), you can handle IPython notebook with 'model.ipynb' . By using dump method in joblib, you can export your model as 'emotion_modle.pkl'.
Before you generating paintings, save your image under "SinGAN/Input/Paint", and run the command
python main.py
- Tamar Rott Shaham, Tali Dekel, Tomer Michaeli "SinGAN: Learning a Generative Model from a Single Natural Image", ICCV 2019
- 'SinGAN' Github Repository (https://github.com/tamarott/SinGAN)