###
:2019,28(2):1-7
本文二维码信息
码上扫一扫!
基于改进损失函数的YOLOv3网络
吕铄1,2,3, 蔡烜4, 冯瑞1,2,3
(1.复旦大学 计算机科学技术学院, 上海 201203;2.上海视频技术与系统工程研究中心, 上海 201203;3.复旦大学 智能信息处理实验室, 上海 201203;4.公安部第三研究所物联网技术研发中心, 上海 201204)
YOLOv3 Network Based on Improved Loss Function
LYU Shuo1,2,3, CAI Xuan4, FENG Rui1,2,3
(1.School of Computer Science, Fudan University, Shanghai 201203, China;2.Shanghai Engineering Research Center for Video Technology and System, Shanghai 201203, China;3.Laboratory of Intelligent Information Processing, Fudan University, Shanghai 201203, China;4.R & D Center of Internet of Things, The Third Research Institute of Ministry of Public Security, Shanghai 201204, China)
摘要
图/表
参考文献
相似文献
本文已被:浏览 1628次   下载 1100
投稿时间:2018-08-12    修订日期:2018-09-05
中文摘要: 为了提高卷积神经网络在目标检测的精度,本文提出了一种基于改进损失函数的YOLOv3网络.该网络模型应用一种新的损失函数Tan-Squared Error (TSE),将原有的平方和损失(Sum Squared Error,SSE)函数进行转化,能更好地计算连续变量的损失;TSE能有效减低Sigmoid函数梯度消失的影响,使模型收敛更加快速.在VOC数据集上的实验结果表明,与原网络模型的表现相比,利用TSE有效提高了检测精度,且收敛更加快速.
Abstract:To improve the object detect precision of Convolutional Neural Network (CNN), we present a YOLOv3 network which based on improved loss function. This network model uses a new loss function Tan-Squared Error (TSE) which transferred from primary Sum Squared Error(SSE), and works better on continuous variable error computing. Meanwhile, the properties of TSE could decrease the impact of vanishing gradient problem in sigmoid function, and speed up model converging. The experiment results in Pascal VOC dataset show that TSE improves the detect precision effectively compared with the performance of primary network model, and the convergence is accelerated.
文章编号:     中图分类号:    文献标志码:
基金项目:国家重点研发计划(2017YFC0803700);上海市科委项目(17511101702);复旦大学工程与应用技术研究院先导项目(gyy2917-003)
引用文本:
吕铄,蔡烜,冯瑞.基于改进损失函数的YOLOv3网络.计算机系统应用,2019,28(2):1-7
LYU Shuo,CAI Xuan,FENG Rui.YOLOv3 Network Based on Improved Loss Function.COMPUTER SYSTEMS APPLICATIONS,2019,28(2):1-7

用微信扫一扫

用微信扫一扫