AI CCTV Analytics Software Development

I need a piece of software that plugs straight into the video stream coming from existing, off-the-shelf CCTV cameras and immediately adds deep-learning smarts. The core must deliver three capabilities with high accuracy: object detection, facial recognition and motion tracking and more

Whenever the system identifies a configured event it should push an alert to me by email, SMS, an in-app popup and, for operators in the control room, an on-screen monitor notification. Ideally the alert payload will include a timestamp, camera ID and a cropped snapshot or short clip of the trigger moment.

Please build the solution so it can run on a local server (GPU-enabled Windows or Linux) but remain flexible enough to shift to a cloud instance later if needed. A clean, web-based dashboard to review live detections, manage watchlists and tune sensitivity would be a big plus.

Deliverables
• Installable software (source + compiled package) that connects to standard RTSP/ONVIF camera feeds
• Model training or transfer-learning pipeline that achieves high accuracy on my sample footage
• Real-time alert module covering email, SMS, in-app and monitor popups
• Setup guide and brief user manual

I’m comfortable if you leverage frameworks such as Python, OpenCV, TensorFlow or YOLO, as long as performance remains near real-time on 1080p streams. Let me know your approach, estimated turnaround time and any prerequisites you’ll need from my side (e.g., labeled footage, server specs).