Skip to content

Demo

BirdBox offers an interactive demo to run model inference on single audio files. The environment is capable of CUDA, has all dependencies pre-installed and can access the most recent BirdBox YOLO-models.

Demo Limitations

The hosted demo runs on an RTX 3080 Ti (12 GB VRAM). Due to this shared hardware, the following hard limits apply:

  • Max. concurrent users: 10
  • Max. file size: 300 MB
  • Max. file length: 10 minutes

For unrestricted use, run the CLI on your own system (see CLI Reference) or host the demo yourself.

Self-hosted demo

To host the demo yourself you have to setup the BirdBox environment beforehand. If the installation is complete just run:

streamlit run src/streamlit/app.py

Interface

If everything works as intended you should see an interface similar to this one:

Streamlit UI

Output Formats

For detailed explanations of the output formats, see Data In/Out.