Monitoring the people count accurately within a closed space is critical for several applications such as crowd management, fire safety and effective space usage. This project resolves this issue by proposing a refined solution through the application of MobilNet-SSD based object detection model with Centroid Tracking algorithm based on OpenCV for tracking the motion of detected individuals between successive frames. This two-way method combining object detection and tracking allows for proper counting of individuals passing in or out of the area under surveillance, even with moderate occlusions and overlapping motion. One of the system’s pre-requisites is a carefully placed camera, right over all the doors of the room under surveillance. A key element to this system is the real time notification service and several features such as scheduling, threading and logging. The main factors for the assessment of this system are accuracy and efficiency. These factors demonstrate how practical the system is for use in real life and also poses an affordable solution with high accuracy. The system provides trade-off between detection quality and operation efficiency without using specialized sensors or sophisticated infrastructure.