摘要
研究了基于谷歌深度学习框架Tensorflow的深度学习在车型识别领域应用的可能性,使用了卷积神经网络模型,将6种不同类型的车辆图片作为训练集输入神经网络,通过多次训练,选择合适的神经网络超参数,最终得到了可以通过输入图片直接识别汽车类型的卷积神经网络模型,且准确率约为92%。同时,通过学习,以Tensorflow为后端,keras为前端进行建模、编程的完整流程,为进一步使用Tensorflow深度学习框架构建图像识别的应用环境打下了基础。
This paper studies the possibility of the application of deep learning based on Google deep learning framework TensorFlow in the field of vehicle recognition.Using the convolutional neural network model,six different types of vehicle pictures are input into the neural network as the training set,and the appropriate neural network super parameters are selected through repeated training.Finally,the convolutional neural network model,which can directly identify vehicle types through input images,is obtained with an accuracy rate of about 92%.At the same time,by learning the complete process of modeling and programming with TensorFlow as the back end and Keras as the front end,it lays a foundation for further constructing the application environment of image recognition with TensorFlow deep learning framework.
作者
何梓林
He Zilin(Shanghai University of Science and Technology,Shanghai 200093,China)
出处
《农业装备与车辆工程》
2022年第8期135-138,共4页
Agricultural Equipment & Vehicle Engineering