Real-Time Abnormal Activity Detection System using Python and Mediapipe with Automated Alert Mechanism
Real-Time Abnormal Activity Detection System using Python and Mediapipe with Automated Alert Mechanism
Dr. Neeta Deshpande¹, Maseera A Sayyed²
¹Associate Professor, Computer Engineering Department, Gokhale Education Society’s,
R. H. Sapat College of Engineering, Management Studies and Research, Nashik,
Affiliated to Savitribai Phule Pune University, Pune, Maharashtra, India
²P.G. Student, Computer Engineering Department, Gokhale Education Society’s,
R. H. Sapat College of Engineering, Management Studies and Research, Nashik,
Affiliated to Savitribai Phule Pune University, Pune, Maharashtra, India
Abstract:— Advancements in computer vision and machine learning have significantly transformed the domain of intelligent surveillance systems. This paper presents a real-time Abnormal Activity Detection System developed using Python and the MediaPipe framework. The proposed system continuously monitors live video streams to identify unusual or suspicious human behaviors by extracting skeletal pose landmarks from video frames. These landmarks are mathematically evaluated against a pre-trained repository of normal gesture patterns using the Euclidean distance metric. Upon detection of an anomalous activity, the system autonomously triggers an alert mechanism that delivers an email notification to designated security personnel, embedding the captured incident image along with the corresponding geographical location. The architecture supports multi-location monitoring through a role-based login interface, an administrative dashboard, and a comprehensive incident logging module. Experimental evaluations demonstrate that the system achieves reliable detection accuracy at a processing rate of 15 to 20 frames per second, with alert delivery latency under five seconds. The framework proves to be computationally lightweight, platform-independent, and readily deployable in environments such as offices, educational institutions, public spaces, and residential areas.Keywords — Abnormal Activity Detection, MediaPipe, Euclidean Distance, Pose Estimation, Human Activity Recognition, Computer Vision, Real-Time Surveillance, Email Alert System, OpenCV, Python.