计算机与数字工程2025，Vol.53Issue(10)：2688-2692,5.DOI:10.3969/j.issn.1672-9722.2025.10.003
基于深度学习的目标检测算法研究
Research on Target Detection Algorithm Based on Deep Learning
摘要
Abstract
This paper improves the generation model generator and discrimination model discriminator of the Generative Ad-versarial Network(eg GAN)network according to the working principle of the GAN network and the characteristics of the Fast R-CNN model.Combining the idea of adding label constraints to Auxiliary Classifier GAN(eg AC-GAN)and Deep Convolutional GAN(eg DC-GAN),and using Convolutional Neural Network(eg CNN)to replace the characteristics of multi-layer perceptron in GAN in both generator and discriminator,this paper proposes Auxiliary Classifier Deep Convolutional GAN(eg AC-DCGAN)mod-el.This model is added with multi-classification and condition auxiliary options and batch normalization operation in the generation model and discrimination model,and uses convolution and deconvolution instead of a pooling layer,and uses global pooling layer instead of a full connection layer.It has been tested on the COCO data set,Pascal VOC 2007 data set,and Pascal VOC 2012 data set,and these tests have achieved good results.关键词
生成对抗网络/卷积神经网络/AC-GAN/DC-GAN/AC-DCGANKey words
generative adversarial network/convolutional neural networks/AC-GAN/DC-GAN/AC-DCGAN分类
信息技术与安全科学引用本文复制引用
LI Shuxia,YANG Juncheng..基于深度学习的目标检测算法研究[J].计算机与数字工程,2025,53(10):2688-2692,5.基金项目
河南省科技厅科技攻关项目(编号:242102210072) （编号:242102210072）
河南省教育厅高等学校重点科研项目(编号:22B520009) （编号:22B520009）
河南工业职业技术学院青年骨干教师培养计划 （）
全国高等院校计算机基础教育研究会纵向课题(编号:2021-AFCEC-202)资助. （编号:2021-AFCEC-202）
