FPGA accelerated abandoned object detection

Abstract

In this paper, a method to address the problem of detecting abandoned object(s) in a live video feed has been presented. The proposed technique utilizes a static background modeling algorithm and identifies any object lying abandoned for a given period of time. Our approach uses a FIFO queue for each pixel in a video frame as an essential element and applies a processing logic. Since the serial processing of such pixel queues on a conventional computing platform is relatively a slow process, the proposed algorithm is synthesized on an FPGA with the aim of making a custom chip that can be embedded in a development board for abandon object detection applications. Such hardware implementation also speeds up the execution of the algorithm by exploiting the parallel nature of algorithm and makes the computation real-time. The proposed technique successfully detects abandoned objects of different sizes present in the video.