针对嵌入式设备的YOLO目标检测算法改进方法
Improvement methods for YOLO object detection algorithm targeting embedded devices
张立国 1孟子杰 1金梅1
作者信息
- 1. 燕山大学电气工程学院 秦皇岛 066000
- 折叠
摘要
针对算法在资源有限的嵌入式设备实现困难的问题,本文基于YOLO系列算法提出适应嵌入式设备实现的轻量化改进方法.方法具体包括:基于YOLOv4-Tiny算法结构,引入GhostNet思想改进其网络主干,大量降低网络参数量和计算量;通过加强颈部网络特征融合效果,减少模型压缩导致的精度损失;采用训练中量化的方式将网络模型参数从32位浮点型数据转换为适合嵌入式设备计算的8位定点型参数.实验结果表明,改进后的网络在检测精度满足应用要求的情况下,模型尺寸相对原算法降低57％,在嵌入式设备上实现功耗仅3.795 W.
Abstract
To address the problem of implementing algorithms on resource-limited embedded devices,a lightweight im-provement is proposed based on the YOLO series of algorithms to adapt to embedded device implementation,specif-ically including:improving the network backbone by introducing GhostNet ideas based on the YOLOv4-Tiny algo-rithm structure to significantly reduce network parameters and computational complexity;strengthening the fusion effect of neck network features to reduce accuracy loss caused by model compression;and using quantization during training to convert network model parameters from 32-bit floating-point data to 8-bit fixed-point parameters suitable for embedded device computation.Experimental results show that after the improvement in this paper,the network's model size relative to the original algorithm is reduced by 57％when the detection accuracy meets application re-quirements,and the power consumption for embedded device implementation is only 3.795 W.
关键词
目标检测/YOLOv4-Tiny/轻量化设计/嵌入式实现/加速器Key words
object detection/YOLOv4-Tiny/lightweight design/embedded implementation/accelerator引用本文复制引用
基金项目
国家重点研发计划(2020YFB1711001)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据