Juan Liu / ChangSha university of science and technology
Lijun Tang / ChangSha university of science and technology
Zhigang Zhang / ChangSha university of science and technology
Longhao Zheng / ChangSha university of science and technology
Feng Bin / ChangSha university of science and technology
Yongjun Wen / ChangSha university of science and technology
In the traffic field, the traffic sign detection and recognition system is an important part of ADAS (Advanced Driver Assistance System). In this paper, a research on traffic sign recognition is carried out, and a traffic recognition sign recognition network is established with the GTSRB data set as the research object. A network computing accelerator is designed with "ARM+FPGA" framework; the functional modules of each part of the network are reasonably divided; and high-level synthesis tools are used to optimize the design of loop unwinding and loop flow to improve throughput and improve recognition speed. The experimental results show that the network has a high accuracy rate in each classification. After optimization, the processing speed of the system under the 100MHz clock is increased by 5.59 times. This method provides an effective reference for the identification of other specific signs.