This is a demonstration of a human detection, tracking and counting system built using deep learning based computer vision. For human detection Pedrestron model has been used. Kalman Filtering and Hungarian algorithm have been used for tracking and data association respectively. It can keep the count of peple who passed a line in each direction.