基于FPGA平台的VGG16网络系统分析  

Analysis of VGG16 Network System Based on FPGA Platform

作　　者：王鑫海 黄世震[1] 林伟峰 WANG Xinhai;HUANG Shizhen;LIN Weifeng(Fuzhou University,Fujian 350108,China.)

机构地区：[1]福州大学,福建350108

出　　处：《电子技术(上海)》2025年第1期62-64,共3页Electronic Technology

摘　　要：阐述随着边缘计算的发展,智能搜索、图像理解、无人机和自动驾驶等边缘场景对对象检测技术提出了更高的要求。为此,基于FPGA平台,在性能和缓存大小之间取得平衡,设计一种可调整资源大小的FPGA实现的VGG16网络。该电路架构有助于用户在不同成本的FPGA上实现深度卷积神经网络模型。实验结果表明,这种卷积神经网络架构优于典型的边缘平台。This paper describes that with the development of edge computing,such as intelligent search,image understanding,unmanned aerial vehicles and autonomous driving,edge scenes need object detection technology.It is based on FPGA platform and balances performance and cache size.It designs a VGG16 network implemented by FPGA that can adjust the size of resources.This circuit architecture helps users implement deep convolutional neural network models on FPGAs of different costs.The experimental results show that this convolutional neural network architecture is superior to typical edgeplatforms.

关 键 词：卷积神经网络 边缘计算 FPGA VGG-16 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]