Multi-pedestrian Assignment and Detection Based on DE-Greedy Algorithm

This article is aimed at an actual project for venue security which means there are dynamic targets and detectors. To solve this problem, we conducted preliminary research and analysis. In order to improve the efficiency of UAV-detector for dynamic targets and minimize the execution time, this article proposes a model using DE (Differential Evolution) algorithm to simulate the moving of individuals and realize task-allocation. We also analysed different situations for evaluation function with high density and low density. Firstly, the initialization, mutation and cross-select of DE optimization algorithm are used to simulate moving individuals and selection is used to calculate cost of each decision. After that, the greedy algorithm is used to sort the execution order for which we choose different parameters in different situations. Usually the way of computer simulation will be adopted to simulate a variety of conditions of traction load and the static simulation is done to verify the evaluation and method.