Search

Video analytics is emerging as a high potential area supplementing intelligent transportation systems (ITSs) with wide ranging applications from traffic flow analysis to surveillance. Object detection and classification, as a sub part of a video analytical system, could potentially help transportation agencies to analyze and respond to traffic incidents in real time, plan for possible future cascading events, or use the classification data to design better roads. This work presents a specialized vehicle classification system for urban environments. The system is targeted at the analysis of vehicles, especially trucks, in urban two lane traffic, to empower local transportation agencies to decide on the road width and thickness. The main thrust is on the accurate classification of the vehicles detected using an evolutionary algorithm. The detector is backed by a differential evolution (DE) based discrete parameter optimizer. The authors show that, though employing DE proves expensive in terms of computational cycles, it measurably improves the accuracy of the classification system.

Differential Evolution (DE) is a highly competitive and powerful real parameter optimizer in the diverse community of evolutionary algorithms. The performance of DE depends largely upon its control parameters and is quite sensitive to their appropriate settings. One of those parameters commonly known as scale factor or F, controls the step size of the vector differentials during the search. During the exploration stage of the search, large step sizes may prove more conducive while during the exploitation stage, smaller step sizes might become favorable. This work proposes a simple and effective technique that alters F in stages, first through random perturbations and then through the application of an annealing schedule. We report the performance of the new variant on 20 benchmark functions of varying complexity and compare it with the classic DE algorithm (DE/Rand/1/bin), two other scale factor altering variants, and state of the art, SaDE.