Ph4n70mC0de/Smart-Surveillance-System ? reverse-engineered prompt

Build me a smart surveillance system for a municipality that works with existing 5MP IP cameras and RTSP streams, without replacing the cameras. I want it to analyze live video, detect incidents with AI, and show the results on a web dashboard with annotated feeds and clear alerts.

Use YOLOv8 with 640 by 640 processing, plus non maximum suppression and a sliding window temporal check so the alerts are not noisy. The system should cover five incident types, support human review before any action, and send SMS alerts through Twilio based on the approved workflow. Please make it privacy focused, so no facial recognition, no biometric identification, and no audio capture.

Build the backend with FastAPI and PostgreSQL, and make the key settings like confidence threshold, temporal window, and cooldown easy to change. Also include the dataset and training workflow, live feed handling, basic admin controls, and evaluation support for the pilot study. If you need to check current docs or best practices online, go ahead and look them up.