tubakhxn/computervision_projects — reverse-engineered prompt

Build me a simple Python computer vision project that matches this repo and actually works end to end. I want three demo modes, one for air combat style aircraft tracking from video, one for license plate and vehicle recognition, and one for airport runway monitoring. Each mode should let me open a video file or camera feed, detect and track the relevant objects, draw clear overlays on screen, and show useful analytics in a basic dashboard or side panel.

For the aircraft demos, include path visualization, activity monitoring, and simple prediction of where tracked aircraft are heading. For the license plate mode, detect vehicles and plates, read the plate text with OCR, score confidence, track vehicles, and export results to CSV. For the airport mode, show runway occupancy, taxiway activity, and basic takeoff and landing detection.

Please keep it educational and safe, not a real weapons system. If the repo is messy, reorganize it into a clean runnable app with one obvious entry point and install any needed dependencies automatically. Look up current docs online if you need to.