- Jinil Jang
- Chi Wang
http://cs230.stanford.edu/projects_spring_2019/reports/18681213.pdf
http://cs230.stanford.edu/projects_spring_2019/posters/18681474.pdf
pip install -r requirements.txt
There are 2 ways to prepare dataset; Download images from google
or Download pre-collected images from S3
cd ~/DeepVideoFaceSwap && mkdir download && aws s3 cp --recursive s3://faceswap-dataset ./download
python faceswap.py collect -o "download" -k "george_clooney" --limit 100
If you want to download images more than 100, you need to install Chrome. Follow this:
- Linux
- Install Chrome:
curl https://intoli.com/install-google-chrome.sh | bash
- Download Chromedriver:
cd ~/DeepVideoFaceSwap && wget https://chromedriver.storage.googleapis.com/75.0.3770.8/chromedriver_linux64.zip && unzip chromedriver_linux64.zip && rm chromedriver_linux64.zip
- Install Chrome:
python faceswap.py collect -o "download" -k "george_clooney" --limit 2000 --driver-path "~/DeepVideoFaceSwap/chromedriver"
python faceswap.py extract -i ./download/images/george_clooney -o ./download/images/george_clooney_out/
python faceswap.py extract -i ./download/images/jinil_jang -o ./download/images/jinil_jang_out/
python faceswap.py train -A ./download/images/george_clooney_out/ -B ./download/images/jinil_jang_out/ -m ./output/model/george_jinil/
python faceswap.py convert -i ./download/images/george_clooney/ -o ./output/converted/george_jinil/ -m ./output/model/george_jinil/