Activity Monitor System is an AI-powered video analytics solution designed to detect, track, and analyze mobile-phone usage in real time from live or recorded video sources. The system uses computer vision to identify people and mobile phones, associate devices with individuals, and measure sustained phone usage through intelligent temporal analysis. An optional facial recognition layer enables the platform to associate detected activity with enrolled individuals and generate personalized activity reports. Advanced behavioral analysis, including gaze direction, talking, drowsiness, and phone movement detection, provides additional context for more accurate activity classification. The solution processes video entirely on-premise, eliminating dependency on cloud APIs and per-request processing costs. GPU acceleration and CPU fallback provide flexible deployment across different hardware environments. With a React dashboard, FastAPI backend, and detailed reporting capabilities, the system demonstrates our expertise in AI development, computer vision, video analytics, automation, and intelligent enterprise solutions.