The ability to reliably detect and track human motion is a useful tool for higher-level applications that rely on visual input. Interacting with humans and understanding their activities are at the core of many problems in intelligent systems, such as human-computer interaction and robotics. Proposed system detect the motion of real world objects with the help of stationary camera and from video using Improved frame subtraction technique. Results of proposed system are evaluated on various sample data collected from real world situations and experiments shows that the accuracy of the proposed system is better than of existing systems.