【摘要】 为了将有效地识别车辆类型用于智慧交通系统,本文在分析Inception V3模型的基础上,提出了一种基于迁移学习理论的车型分类深度学习模型。该模型首先在Inception V3模型的基础上去除最后的全连接层,并加入参数优化层,然后采用Dropout和全局平均池化层。理论分析和试验结果表明,该模型的性能优于基于VGG-16的车型分类模型、基于Xception的车型分类模型和基于Resnet50的车型分类模型,其训练精度优于96.48%、测试精度优于83.86%。更多还原

【Abstract】 In order to incorporate the recognition of vehicle type into the intelligent transportation system, a deep learning model for vehicle classification based on transfer learning theory was proposed in terms of the analysis of the Inception V3 model. The model removes the last fully connected layer based on the Inception V3 model, and adds a parameter optimization layer, and then uses Dropout and the global average pooling layer. Theoretical analysis and experimental results show that the performance of the model is better than the vehicle classification models based on VGG-16, Xception and Resnet50, with a training accuracy above 96.48% and a testing accuracy above 83.86%.更多还原