The job is more than reading numbers
An automatic score follower needs stable camera input, scoreboard region detection, OCR or custom ML for digits, confidence checks, event timing, and a fallback path when glare, motion, LED refresh, or camera angle makes a frame unreliable.
A realistic edge-to-cloud pipeline
Jetson Nano can handle local video ingest, GStreamer pipelines, OpenCV preprocessing, and lightweight model inference. AWS Media Live can support broadcast-style streaming while score events, timestamps, and overlays are generated from the edge pipeline.
Useful features for schools and local sports
The system can track score changes, keep event logs, support live overlays, mark clips around score events, detect scoreboard visibility problems, and send operator alerts when confidence is low. A human can approve corrections without stopping the stream.