The Design And Implementation Of The Vehicle Type Classification System Based On Convolutional Neural Network

With the continuous development of rising social and economic conditions, the vehicle population has increased. The main functions of intelligent transportation system are to accurately detect vehicle information and the correct identification of passing vehicles. The current vehicle detection and classification technology are in two ways:Auto vehicle detection and Auto vehicle classification. Faced with such massive dynamic traffic data, how efficient the collection, transmission, rapid analysis, in order to effectively manage the public security organs of law and order situation, is a very big challenge. The efficiency of traditional vehicle classification tools is slow, most of the known vehicle classification is limited to large, medium and small category, they cannot achieve fine classification and the newly produced vehicle models cannot be quickly updated, it has been unable to meet large data norm at high efficiency and high accuracy the analysis needs, in big data context, how will high quality and efficient application of new technologies to the processing of traffic data is a problem worthy of study and reflection, our vehicle classification system is based on this background design and development of.Appears deep learning theory and technology to solve the above problems provides a good solution ideas. We use this theory as applied to image processing and classification of the theoretical model into the data processing, the use of "convolution neural networks" to construct the network structure model,using calculating framework Caffe to handle traffic data and use distributed message queue to push the analysis results to the front-end interface in real-time. We designed and implemented the whole process.System is based on MVC structure, using SpringMVC as a system basis framework. According to the needs of users, we accomplished the system demand analysis, outline design, detailed design of the module, and after completion of the development we had a comprehensive testing to ensure the stability of the system.