Research on Vehicle Identification Technology Based on SOPC

Haoran Li ； Peng Zhang ； Yan Li

《International Core Journal of Engineering》 7卷3期 (2021/03) Pp. 109-115

https://doi.org/10.6919/ICJE.202103_7(3).0016

摘要

In recent years, with the rapid increase of computing power and interpenetration and continuous development of many subjects, artificial intelligence has gradually entered the public's vision. Deep learning is an important means of artificial intelligence. This paper mainly studied and implemented how to apply the technology of deep learning on the System-on-a-Programmable-Chip platform to complete the function of recognizing and classifying the vehicle image data collected by the camera. The main content of the paper is as follows. Firstly, it introduced the knowledge related to deep learning and determined the CIFAR-10 data set. Based on this, the construction and training of Convolutional Neural Networks under the framework of caffe network and the way to use external data to test the reliability of the network were been realized. And the network reliability test and the extraction of important parameters of the convolutional neural network were performed. Next, a hardware platform based on SOPC-based camera data acquisition and display is designed. The platform provides hardware support for forward recognition algorithms for convolutional neural networks. At the same time, the forward recognition algorithm for convolutional neural networks was implemented by using C language on the operating system side. After this, by using the accelerator technology of the SDSOC development platform, the C language functions that affect the speed of the system was converted into the hardware circuit of the FPGA, and the entire forward identification algorithm was accelerated. Finally, the entire system was tested and the experimental and the results were analyzed to confirm that the system can achieve the desired function.

關鍵字

SOPC ； Vehicle Identification ； Deep Learning ； Convolutional Neural Network ； Accelerator

參考文獻

LeCun Y, Bengio Y, Hinton G.2015. Deep learning[J]. Nature: 436-444.

Huang Linan. Research and implementation of graph data processing system based on FPGA [D]. 2016.

Liu Hanying, Zhang Yaotian, Zhang Yuxi, et al. SAR Target Recognition Based on deep learning and FPGA implementation [C]//11th National Symposium on signal and intelligent information processing and application.

Wang Jia (2014) accelerator research of deep learning [D]. University of Chinese Academy of Sciences.

Xu Guangqiang. 2017. Design of single image recognition algorithm based on FPGA [D]. Heilongjiang University.