A minimalist implementation of a video labeling interface. We internally use this tool to annotate videos for cleanliness in order to train a data filtering model (cf. Video Pre-training, Genie).
Install the requirements with:
pip install -r requirements.txtNext, set your video folder, a database URL, and the tags of your choice in a .env file:
# ./.env
VIDEO_FOLDER_PATH="./data"
DATABASE_URL="sqlite:///./vid_tag.db"
PREDEFINED_TAGS='["TagA", "TagB", "TagC", "Needs Review"]' Finally, simply run the webserver:
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000You can annotate by clicking on the tags or by using the provided keybindings (number keys by default).
vid-tag saves your annotations in a lightweight sqlite3 database with the following table schema:
| id | filepath | filename | is_annotated | tags |