基于硬件描述语言的目标识别硬件加速器设计  

Design of object recognition hardware accelerator based on hardware description language

作　　者：张嘉 金婕[1] ZHANG Jia;JIN Jie(College of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《智能计算机与应用》2023年第6期108-111,共4页Intelligent Computer and Applications

基　　金：国家自然科学基金(61801286)。

摘　　要：本文针对深度学习方法在目标识别领域内的应用,设计了一种基于硬件描述语言的目标识别硬件加速器,运用数据流架构优化方法,设计了二值化卷积神经网络算法所对应的硬件模块单元,实现了对输入图片的识别。实验结果表明,基于多帧分辨率为224×224的图片输入,在硬件平台仿真软件中达到了不俗的识别速率以及识别准确率,为基于硬件系统的深度学习加速器的研究奠定了基础。Aiming at the application of deep learning method in the field of target recognition,this paper designs a hardware accelerator for target recognition based on hardware description language.This paper uses the data flow architecture optimization method to design the hardware module unit corresponding to the binarized convolutional neural network algorithm,and realizes the recognition of the input image.The experimental results show that under the input of multi-frame images with a resolution of 224×224,excellent recognition speed and recognition accuracy are achieved in the hardware platform simulation software,which lays the foundation for the research of deep learning accelerators based on hardware systems.

关 键 词：硬件描述语言 二值化卷积神经网络 深度学习 硬件加速器 

分 类 号：TP332[自动化与计算机技术—计算机系统结构]