计算机与数字工程2024，Vol.52Issue(4)：1103-1109,7.DOI:10.3969/j.issn.1672-9722.2024.04.025
基于改进Faster R-CNN的多种类车灯检测方法
Multi-type Auto Lamp Detection Method Based on Improved Faster R-CNN
摘要
Abstract
Autonomous spray painting robots can realize the automatic spraying of various types of automobile lamps,and the lamp detection algorithm based on machine vision is the key technology of the robot.In view of the lack of deep learning algorithm to detect auto lamps in spraying environment,a detection algorithm based on improved Faster R-CNN is proposed.In the improved al-gorithm,the dense residual network(De-ResNet)is used to replace the original feature extraction network,which integrates multi-level feature information,increases the network depth,and avoids the disappearance of network gradient.At the same time,Distance-IoU(DIoU)is used to improve the loss function in the original algorithm.The improved IOU introduces a good distance measurement to further improve the detection accuracy.The experimental results show that the average accuracy of the improved al-gorithm is 98.56%,and the average recognition time of a single image is 0.45 s.It can realize the effective recognition of lamp types and meet the requirements of real-time processing.关键词
喷涂机器人/目标检测/Faster R-CNN/密集连接网络/损失函数Key words
spray painting robot/object detection/Faster R-CNN/DenseNet/loss function分类
信息技术与安全科学引用本文复制引用
郭碧宇,陈伟,张境锋,魏庆宇..基于改进Faster R-CNN的多种类车灯检测方法[J].计算机与数字工程,2024,52(4):1103-1109,7.基金项目
镇江市国际科技合作项目(编号:GJ2020009) （编号:GJ2020009）
镇江市产业前瞻与共性关键技术前期引导项目(编号:GY2018018)资助. （编号:GY2018018）
