Identification and Localization of Flexible Arm End Based on DS-Faster R-CNN Algorithm
RU Changxin
ZANG Hongbin
DENG Xiandong
YANG Huailin
Abstract：Aiming at the problems of Faster R-CNN target detection algorithm in vision robots,such as weak generalization ability and large consumption of computing resources,a DS-Faster R-CNN algorithm based on improved DenseNet network is pro-posed.Firstly,channel attention ECA is fused at the feature network layer.The data set is constructed by K-means clustering to gen-erate anchor boxes,and different optimization schemes are used to improve the learning and expression ability of the model.The av-erage accuracy of the improved algorithm(mAP)is increased by 1.6%.Secondly,Siamese-RPN tracking algorithm is integrated to obtain the three-dimensional coordinates of the end of the flexible arm to guide the movement of the flexible arm and improve the fi-nal positioning accuracy.
Keywords：image processingobject detectionconvolutional networkfeature extractiontarget trackingflexible arm
Publication Date：2025-10-20
Online Publishing Date：2026-01-16(First online date of this platform, not the publication date of the document)
Pages：8( 2779-2786 )