一种综合赋权的算法性能评价体系  

A Comprehensive Weighted Algorithm Performance Evaluation System

作　　者：许星来 高洋[2] 明德烈[1] XU Xinglai;GAO Yang;MING Delie(National Key Laboratory of Multispectral Information Processing Technology,School of Artificial Intelligence and Automation,Huazhong University of Science and Technology,Wuhan 430074;Beijing Aerospace Automatic Control Institute,Beijing 100854)

机构地区：[1]华中科技大学人工智能与自动化学院多谱信息处理技术国家级重点实验室,武汉430074 [2]北京航天自动控制研究所,北京100854

出　　处：《计算机与数字工程》2024年第2期327-331,共5页Computer & Digital Engineering

摘　　要：针对当前自动目标识别算法的性能评价指标过于单一的问题,以行人检测作为应用背景下进行了相关研究。论文设计了一套基于AHP(层次分析法)与CRITIC赋权法的目标识别算法性能评价指标体系。该指标体系针对目标识别算法的准确性、时效性、适应性设计了多个性能评价指标,并综合主观赋权法与客观赋权法对评价指标进行赋权。实验测试结果与ImageNet等计算机视觉挑战赛中person类别的检测排名基本吻合,证实论文提出的性能评价体系可靠。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 adaptability 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.

关 键 词：自动目标识别 性能评价体系 多属性指标 综合赋权 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]