Real Time Dart Detection System

I’m building an automated scorer for a standard dartboard that relies on three off-the-shelf webcams. As soon as a dart sticks, the system must pinpoint its exact segment, translate that into the correct score, and push a real-time event—ideally over a WebSocket—so any client can update the match instantly.

Core expectations
• Camera input: three synchronized 1080p webcams positioned around the board.
• Tech stack: feel free to reach for either Python/OpenCV or C++/OpenCV; whichever lets you reach millisecond-level detection with reliable accuracy.
• Calibration: a quick routine that someone in a pub can run in a couple of minutes—no chessboard targets or lab lighting required.
• Output: headless service only; no on-screen UI. Emit JSON events such as `{ "x":…, "y":…, "ring":"double", "number":20, "score":40 }` the moment impact is confirmed.

Deliverables
1. Source code with build/run instructions.
2. Calibration workflow and any printable targets if you use them.
3. API spec for the real-time event stream plus a minimal test client.
4. Performance report: detection accuracy, average latency, and test footage.

Acceptance criteria
• ≥98 % hit-segment accuracy across the whole board.
• End-to-end latency ≤150 ms on a mid-range PC.
• Calibration completed in ≤2 minutes by a first-time user.

If this sounds like your kind of challenge, let’s talk through your approach to multi-camera geometry, lighting variance, and fast image processing so we can get darts flying and scores flowing.