摘要
本文提出了一种基于改进残差网络ResNet50模型的图像识别方法。通过引入圆形LBP算法,提取图像内部的纹理信息构成纹理图谱;同时在网络中加入通道注意力机制(Efficient Channel Attention,ECA)提高模型性能,使得改进后的算法更适合识别交通标志。该方法在GTSRB和BelgiumTS交通标志数据集上的准确率分别达到99.7%和98.3%,有效提高了智能系统识别交通标志的准确率和驾驶的安全性。
The correct recognition of traffic signs is the key technology of intelligent driving and unmanned driving.This paper proposes an image recognition method based on the modified residual network ResNet50 model.The circular LBP algorithm is introduced to extract the texture information inside the image to form the texture map.The Efficient Channel Attention(ECA)mechanism is added to the network to improve the performance of the model,making the improved algorithm more suitable for recognizing traffic signs.The accuracy rate on the GTSRB and BelgiumTS traffic sign datasets reached 99.7%and 98.3%,respectively,effectively improving the accuracy of recognizing traffic signs and driving safety in the intelligent system.
作者
傅融
彭淼
逯洋
FU Rong;PENG Miao;LU Yang(College of Mathematics and Computer,Jilin Normal University,Siping 136000,Jilin,China)
出处
《智能计算机与应用》
2024年第5期221-226,共6页
Intelligent Computer and Applications
基金
吉林省创新创业人才基金(2023QN31)
吉林省自然科学基金(YDZJ202301ZYTS157)
吉林省发展和改革委员会创新项目(2021C038-7)。
