Integrasi Notifikasi CCTV Melalui Telegram Berbasis Iot Dengan Metode Pose Estimation

This research presents the development of an Internet of Things (IoT)-based automatic notification system integrated with conventional Closed-Circuit Television (CCTV) cameras using the pose estimation method to detect human presence in real-time. The method used in this research is BlazePose from the MediaPipe library, which is capable of detecting up to 33 key points of the human body. The system was developed using the Python programming language with the Flask framework for the backend and Vue.js for the frontend. Video streaming is captured via the Real-Time Streaming Protocol (RTSP) protocol from the CCTV DVR, then analysed using a body pose detection algorithm. If the system detects human presence based on distance, angle, and pose similarity thresholds (cosine similarity), the system will send an automatic notification via Telegram to the user. The test results show that the system successfully detects human poses with a good level of accuracy and can send notifications efficiently according to the specified schedule. The system also provides a web-based feature that displays activity statistics, the number of notifications, and log reports that simplify the monitoring process. This research shows that pose detection technology can be effectively implemented in conventional CCTV systems to improve security responses in real-time and cost-effectively.