Deep Learning Based Vehicle Intelligent Identification Method Research And Implementation

With the continuous growth of the number of vehicles in recent years,the pressure on vehicle management has been increasing.In order to improve the efficiency of vehicle management and enhance the level of autonomy and intelligence of the vehicle management process,there is an urgent need to carry out research on intelligent vehicle recognition methods.Image recognition technology in artificial intelligence technology is an important technology in the information age.Image recognition technology identifies targets through processing and analysis,and its powerful computing power and recognition capability is suitable for large-scale image processing.The combination of vehicle management and artificial intelligence technology will effectively improve efficiency and greatly reduce manpower consumption,which is of great practical significance and application value.This paper mainly proposes adaptive ranks extraction image compression algorithm and image cropping algorithm based on area classification,focusing on vehicle image data acquisition,pre-processing and deep learning based vehicle recognition technology to carry out research,for the problem of vehicle intelligent recognition in daily life environment,design deep learning based vehicle intelligent recognition system,mainly contains: FPGA-based vehicle image acquisition and based on the upper computer The system consists of two parts: FPGA-based vehicle image acquisition and upper computer-based vehicle recognition.In the FPGA part,the adaptive rank extraction image compression algorithm is proposed.Firstly,a camera is used to capture the vehicle image,and an image compression mechanism is established for rank extraction.The vehicle image is extracted according to the requirements of the host computer to achieve image compression,and then the image data is packaged and transmitted to the host computer via Ethernet.In the host computer,the received vehicle images are first pre-processed,and an image cropping algorithm based on region classification is proposed to crop the non-existent vehicle regions and remove the redundant image data,and the YOLOv4 algorithm with a fused attention mechanism is used to achieve the vehicle intelligent recognition function.In order to verify the reliability and applicability of the deep learning-based vehicle intelligent recognition,the core process is simulated and analysed in a semi-physical environment.The simulation results show that this paper successfully implements the FPGA-based image acquisition,compression and transmission functions,and successfully realises the accurate recognition of vehicles in the host computer.