Skip to content

At present, just an example to show how to map the detection algorithm YOLOv2 from model to FPGA

License

Notifications You must be signed in to change notification settings

Ajfil/Xilinx_FPGA_HLS-Mapping-Neural-Network-to-Hardware

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 

Repository files navigation

示例:如何将目标检测算法YOLOv2映射到Xilinx FPGA Zedboard

由于目前处于Deadline时期,要等下个礼拜才有时间把完整的整个流程在Github上分享一下,现在只能先用中文简单描述一下整个流程:从Darknet 框架中的YOLOv2模型到最终生成bitstream的过程。
(1)需要先从Darknet官网下载darknet框架,然后将其中YOLOv2的权重提取出来。由于YOLOv2的卷积层基本都会跟着BatchNormalization,所以在提取权重的过程中,就将YOLOv2中的weight和bias分别和对应的BN乘在一起具体公式可以参考文献[1],公式如下:
y=(x-bn0)/sqrt(bn1) (1)
z=sc0y+sc1 (2)
所以,可以将(1)与(2)合并,得到:
y=A
x+B (3)
其中 A= sc0/sqpr(bn1), B=sc1-sc0*bn0/sqrt(bn1)
这部分代码放在framework目录下

参考文献:
[1] An Automatic RTL Compiler for High-Throughput FPGA Implementation of Diverse Deep Convolutional Neural Networks

About

At present, just an example to show how to map the detection algorithm YOLOv2 from model to FPGA

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published