数据驱动的智能车辆目标检测能力测试评价方法

Data Driven Test and Evaluation Method for Intelligent Vehicle Object Detection Capability

本文针对智能车辆目标检测能力测评存在的指标体系不完整、量化程度和测评实时性低等问题,聚焦智能车辆目标检测能力中的目标分类和目标识别,在这两个测评项目上提出了一套量化的评价指标体系,并用TOPSIS方法进行综合的评价.然后在此指标体系的基础上搭建数据驱动的智能车辆目标检测能力测评平台,平台可满足对智能车辆目标检测能力测评的实时性要求.最后采用了若干组车辆检测算法对指标体系进行验证.

To address the problems such as incomplete index system, low degree of quantification and real-time evaluation of test and evaluation for intelligent vehicle object detection capability, we put forward a set of quantitative evaluation index system on object classification and object recognition. Then, we conduct a comprehensive evaluation by TOPSIS. According to the index system, we establish the data driven platform for intelligent vehicle object detection ability evaluation, the platform can also meet the real-time requirements of the evaluation on intelligent vehicle object detection ability. Finally, several groups of vehicle detection algorithms are used to verify the index system.