Intelligent target detection algorithm for embedded FPGA

CHENG Yu-qing; HE Zhan-zhuang; MA Zhong; BI Rui-xing; MAO Yuan-hong

Chinese

您当前的位置：

首页 >

文章列表页 >

Intelligent target detection algorithm for embedded FPGA

更新时间：2026-09-04

- Intelligent target detection algorithm for embedded FPGA
- Vol. 38, Issue 6, Pages: 87-92(2021)
- 作者机构：
  
  西安微电子技术研究所,陕西,西安,710054
- 作者简介：
- 基金信息：
- DOI：
  CLC：
- Published：2021
- 稿件说明：
移动端阅览
程钰清, 贺占庄, 马钟, 毕瑞星, 毛远宏. 面向嵌入式FPGA的智能目标检测算法[J]. 微电子学与计算机, 2021, 38(6): 87-92.

CHENG Yu-qing, HE Zhan-zhuang, MA Zhong, et al. Intelligent target detection algorithm for embedded FPGA[J]. 2021, 38(6): 87-92.
程钰清, 贺占庄, 马钟, 毕瑞星, 毛远宏. 面向嵌入式FPGA的智能目标检测算法[J]. 微电子学与计算机, 2021, 38(6): 87-92. DOI：

CHENG Yu-qing, HE Zhan-zhuang, MA Zhong, et al. Intelligent target detection algorithm for embedded FPGA[J]. 2021, 38(6): 87-92. DOI：

摘要

随着识别率和实时性的提高，卷积神经网络目标检测算法的计算复杂度和内存需求急剧增加，难以应用在小尺寸和低功耗的嵌入式平台上.本文在分析现有目标检测神经网络模型结构的基础上，根据FPGA高实时性、低功耗以及并行处理的特点，提出了一种在FPGA上高速运算的神经网络模型规整化方法.在此方法指导下设计改进了一款目标检测神经网络模型结构，包括删除LRN层、Scale层的融合和替换Leaky-ReLU为ReLU.通过在voc2007数据集上的对比实验验证了算法结构的有效性，在PC上其速度相比传统YOLO-V1算法提升了11.5%.在Xilinx ZCU102开发板上的仿真表明：该改进的目标检测算法速度达到29 FPS(Frames Per Second)，精度达到62.3 mAP.

Abstract

With the improvement of recognition rate and real-time performance

the computational complexity and memory requirements of convolutional neural network target detection algorithm increase sharply

which makes it difficult to be applied to embedded platform with small size and low power consumption. In this paper

based on the analysis of the existing neural network model structure of target detection

according to the characteristics of high real-time performance

low power consumption and parallel processing of FPGA

a neural network model normalization method based on high speed operation on FPGA is proposed. Under the guidance of this method

a target detection neural network model structure is designed and implemented

including removing the LRN layer

fusion of Scale layer and replacing Leaky-ReLU with ReLU. The effectiveness of the proposed algorithm structure is verified by comparative experiments on VOC2007 dataset. Compared with traditional YOLO-V1 algorithm

the speed of the proposed algorithm on PC is improved by 11.5%. Hardware simulation under Xilinx ZCU102 shows that the improved target detection algorithm can reach the speed of 29FPS (Frames Per Second) and the accuracy of 62.3mAP.

关键词

Keywords

references

陈辰, 严伟, 夏珺, 等. 基于FPGA的深度学习目标检测系统的设计与实现[J]. 电子技术应用, 2019, 45(8): 40-43. DOI: 10.16157/j.issn.0258-7998.190318. CHEN C, YAN W, XIA J, et al. Design and implementation of FPGA-based deep learning object detection system[J]. Application of Electronic Technique, 2019, 45(8): 40-43. DOI: 10.16157/j.issn.0258-7998.190318.

REDMON J, FARHADI A. YOLO9000: better, faster, stronger[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 6517-6525. DOI: 10.1109/CVPR.2017.690.

GIRSHICK R, DONAHUE J, DARRELL T, et al. Region-based convolutional networks for accurate object detection and segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(1): 142-158. DOI: 10.1109/TPAMI.2015.2437384.

SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv: 1409.1556, 2015.

JIANG Z Q, HUYNH D Q. Multiple pedestrian tracking from monocular videos in an interacting multiple model framework[J]. IEEE Transactions on Image Processing, 2018, 27(3): 1361-1375. DOI: 10.1109/TIP.2017.2779856.

QIU J T, WANG J, YAO S, et al. Going deeper with embedded FPGA platform for convolutional neural network[C]//Proceedings of 2016 ACM/SIGDA International Symposium on Field-Programmable Gate Arrays. Monterey: ACM, 2016: 26-35. DOI: 10.1145/2847263.2847265.

方睿, 刘加贺, 薛志辉, 等. 卷积神经网络的FPGA并行加速方案设计[J]. 计算机工程与应用, 2015, 51(8): 32-36. DOI: 10.3778/j.issn.1002-8331.1405-0335. FANG R, LIU J H, XUE Z H, et al. FPGA-based design for convolution neural network[J]. Computer Engineering and Applications, 2015, 51(8): 32-36. DOI: 10.3778/j.issn.1002-8331.1405-0335.

HAO Z K, LIU Y, QIN H W, et al. Scale-aware face detection[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 1913-1922. DOI: 10.1109/CVPR.2017.207.

王巍, 周凯利, 王伊昌, 等. 卷积神经网络(CNN)算法的FPGA并行结构设计[J]. 微电子学与计算机, 2019, 36(4): 57-62. DOI: 10.19304/j.cnki.issn1000-7180.2019.04.012. WANG W, ZHOU K L, WANG Y C, et al. FPGA parallel structure design of Convolutional Neural Network (CNN) algorithm[J]. Microelectronics % Computer, 2019, 36(4): 57-62. DOI: 10.19304/j.cnki.issn1000-7180.2019.04.012.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Design of a shared-memory-based spaceborne data switching architecture

The linear leastsquares positioning algorithm based on FPGA

Algorithm optimization and hardware acceleration for YOLO post processing

Optimization and implementation of FPGA-based H.264 variable block motion estimation algorithm

A secure encryption authentication system based on integrated structure of TRNG and PUF

Related Author

WANG Runze

TANG Lei

KUANG Nailiang

LIANG Yong

FENG Zhouwu

CUI Mohan

CHEN Tao

ZHU Yifei

Related Institution

Xi'an Microelectronics Technology Institute

School of Electronics and Communication Engineering, East China Normal University

School of Microelectronics, University of Science and Technology of China

Xi'an Institute of Microelectronics Technology

School of Physics and Electronic Engineering, Shanxi University

AI问答

Postal code：100079
Tel：（010）53879206 Email：tmw@bjxintong.com.cn
Technical support is provided by Beijing Founder electronics co., LTD 京ICP备09082226号-64 京公网安备11010602201714号
It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.

⁰