摘要
阐述交通标志中感兴趣区域的提取与卷积神经网络相结合,突出图像中的感兴趣区域,训练了一个卷积神经网络模型,使用更深层的网络模型,引入提前停止的理念,结合GTSRB数据集进行训练。结果表明,网络模型的训练速度加快,准确率较高。
This paper describes the combination of the extraction of the region of interest in traffic signs and Convolutional neural network, highlights the region of interest in the image, trains a Convolutional neural network model, uses a deeper network model, introduces the concept of Early stopping, and combines the GTSRB dataset for training. The results indicate that the training speed of the network model is faster and the accuracy is higher.
作者
付晶
山珍
FU Jing;SHAN Zhen(Shaanxi Institute of Technology,Shaanxi 710300,China)
出处
《集成电路应用》
2023年第7期126-127,共2页
Application of IC
关键词
图像识别
神经网络
交通标志
image recognition
neural network
traffic signs
