📷
Features
Live Object Detection
Uses SSD MobileNet V2 via TensorFlow Hub to detect objects in real-time at ~10 FPS on CPU. Draws bounding boxes and labels on the video stream.
Head Tracking
The robot smoothly tracks the largest detected object, turning its head to follow it across the camera frame.
Antenna Reactions
Antennas wiggle when a new type of object is spotted for the first time — a fun visual cue for discovery.
Web Dashboard
Live annotated video feed with detection list, FPS counter, and a toggle to enable/disable head tracking.
How to Install
1
From the Dashboard
Open the Reachy Mini dashboard, go to Apps, and click Install on this app.
2
Or install manually
pip install git+https://huggingface.co/spaces/backtoengineering/reachy_mini_object_detector
3
Run it
Start the app from the dashboard or run it directly. The web UI is available at http://0.0.0.0:8042