Software Engineering Process for Developing a Person Re-identification Framework

Abstract

In video surveillance, different security spots are considered to be placed in unconstrained locations. Cameras located on corridors, gateways and roads, have raised the need for new video surveillance systems to support activities like monitoring and look and find lost individuals. Several tools from image and video processing, as well as machine learning research areas have been conceptualized, and tested on limited data. However, implementing academic knowledge into full applications following state of the art software engineering techniques is complex. Similarly, small changes theorized in academia may represent big changes in software development. In this paper, a combination of the COMET (Collaborative Object Modeling and Architectural Design Method) and OCEP (Open Community Engagement Model) methods is proposed to develop a person re-identification framework named Perception. The three operational modes implemented in the framework (evaluation, registration and operation) provide flexibility for researchers and developers, as evidenced by the implementation of two application-specific prototypes.