2nd place DataHack winner 🥈, September 2019
VideoThat will automagically generate the perfect video compilation that fits impeccably to the theme song of your choice. The algorithm’s default choice is to create a video clip from Taylor Swift’s videos to an Arctic Monkey song.
This project was created as part of DataHack 2019 Israel, read all about it in our blog post 👾
Ubuntu/debian
sudo apt install ffmpeg
macOS
brew install ffmpeg
Install requirements file
pip3 install -r requirements.txt
Install the newest version of youtube-dl
pip3 install --upgrade youtube-dl
To start downoading all of the videoclips and create a data-base, open the terminal go to the repository location and write
note: the following command will download ~1.5GB to your directory and can take a while
./download-db.sh
Detect all of the scene cuts from the db you just created
./scenedetect_all.sh
Run create_scene_db.py
to create a .pkl
df unified file of all the scenes from the db (this creates scene_db.pkl
and song_db.pkl
)
python3 create_scene_db.py
Run choose_scenes.py
to run the algorithm which chooses taylor swift scenes from the db to match to a song (creates chosen_scenes.pkl
)
python3 choose_scenes.py
Run build_video.py
to create the video file from the chosen scenes
python3 build_video.py
That's Us
Dalya Gartzman GitHub LinkedIn