Performance Improvement Strategies for Person Re Identification over Camera Network

Abstract

In the present age, Video surveillance usage is increased in public places as well as private places for security reasons. Attempts are being made to make the surveillance smarter which could recognize or track people on its own without human intervention. As a step towards the automation, person re-identification is a task of searching a person of interest across the volume of the data obtained from non-overlapping video cameras. More specifically Person re-identification is a task which is used to search the particular person captured in one camera vision from set of images of persons captured in the other camera visions over a span of time. The traditional methods of identifying a person such as face recognition fails and so, the person re-identification involves exploiting other visual features like clothing colour. Also, images of the same person may vary significantly because of variation in illumination, viewpoint, background clutter, scale, resolution and occlusion newline