A Comprehensive Weighted Algorithm Performance Evaluation System
Aiming at the problem that the performance evaluation index of the current automatic target recognition algorithm is too single,related researches have been carried out with pedestrian detection as the application background.This paper designs a set of target recognition algorithm performance evaluation index system based on AHP(Analytic Hierarchy Process)and CRITIC weighting method.This indicator system designs multiple performance evaluation indicators for the accuracy,timeliness,and adapt-ability of target recognition algorithms,and combines subjective weighting and objective weighting methods to weight the evaluation indicators.The experimental test results are basically consistent with the detection rankings of the person category in ImageNet and other computer vision challenges,confirming that the performance evaluation system proposed in this paper is reliable.
国家科技期刊平台
NSTL
万方数据
