NikhilSai-4409/DRS — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a cricket DRS decision review prototype that I can run locally for testing. I want it to take live camera feeds or uploaded cricket delivery videos, detect and track the ball, sync multiple camera angles, show replays, estimate LBW outcomes, and include UltraEdge style audio analysis and a HotSpot style visual simulation.

Please make it clear when the system is not confident. If the model, calibration, tracking, sync, replay speed, or confidence checks are not good enough, it should show REVIEW INCONCLUSIVE instead of pretending it knows.

I also want a simple operator dashboard and a web testing page where I can upload one or two demo videos and see ball tracking, speed, replay, and the decision result. Include an offline demo path with synthetic videos so I can try it without real cameras.

Use Python for the backend, with a local API and live updates, and a desktop or web dashboard for the interface. Add setup instructions, environment config, and basic tests. Look up current docs online if you need to.

Want more depth? Deep Reverse