A Custom based IP Design for object Tracking SoC

This paper involves the implementation of a hardware-software subsystem for object detection and tracking on FPGA. Here we discuss our design of a system to detect and track the object which is in motion. We have integrated a system with existing IPs (Intellectual Property) and designed IP to realize a system for object detection and tracking. From a video-input, the sub-system detects the object of interest and the centroid tracking algorithm is used to track the object. The objective is to reduce resource utilization in a system that uses a simple yet efficient algorithm to analyse moving object's behaviour based on video stream analysis. We have tested the system in an application for human fall detection; to identify human in the image, calculate some parameters of its silhouette shape and analyse them in real-time.