| 447 | 9 | 642 |
| 下载次数 | 被引频次 | 阅读次数 |
提出一种基于YOLOv3的改进算法，用于对进入机场停机坪人员的合法性进行自动识别。首先，结合Ghost-Net网络对YOLOv3的特征提取网络进行优化，将卷积层和归一化层进行合并，减少了参数量，使得模型在嵌入式设备中完成检测任务；其次，在网络中添加了注意力机制，并采用自底向上与反卷积特征融合的方式提升小目标的检测能力，对损失函数和模型参数进行改进，提升了网络的训练和检测效果。实验结果表明，与原始算法相比，所提方法可以减少模型的参数量，提升模型在复杂环境下的检测效果。
Abstract：An improved algorithm based on YOLOv3 to automatically recognize the legitimacy of people entering the airport apron was proposed. Firstly, by combining the GhostNet network to optimize the feature extraction network of YOLOv3, the convolutional layer and the normalization layer were combined to reduce the amount of parameters, so that the model could complete the detection task in the embedded device. Secondly, by adding an attention mechanism in the network and using a bottom-up and deconvolution feature fusion method, the detection ability of small targets was optimized. The loss function and model parameters were improved, and the training and detection effects of the network were improved. The results showed that the proposed method could reduce the amount of model parameters and improve the detection effect in complex environments compared with the original algorithm.
[1] 简海云，林晓蓉，和艳，等.论民用航空与内陆边远城镇的互动发展：以云南为例[J].现代城市研究，2018,33(8):102-108.JIAN H Y,LIN X R,HE Y,et al.The interactive development of civil aviation and in land remote towns:taking Yunnan as an example[J].Modern urban research,2018,33(8):102-108.
[2] REN S Q,HE K M,GIRSHICK R,et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE transactions on pattern analysis and machine intelligence,2017,39(6):1137-1149.
[3] REDMON J,FARHADI A.YOLOv3:an incremental improvement [EB/OL].[2021-05-01].https://www.researchgate.net/publication/324387691.
[4] 佘颢，吴伶，单鲁泉.基于SSD网络模型改进的水稻害虫识别方法[J].郑州大学学报(理学版),2020,52(3):49-54.SHE H,WU L,SHAN L Q.Improved rice pest recognition based on SSD network model[J].Journal of Zhengzhou university (natural science edition),2020,52(3):49-54.
[5] 蒋慧琴，王博霖，马岭，等.一种双视图信息融合的乳腺肿块自动检测算法[J].郑州大学学报(理学版),2020,52(4):28-36.JIANG H Q,WANG B L,MA L,et al.An automatic breast mass detection algorithm with dual-view information fusion[J].Journal of Zhengzhou university (natural science edition),2020,52(4):28-36.
[6] HAN K,WANG Y H,TIAN Q,et al.GhostNet:more features from cheap operations[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.Piscataway:IEEE Press,2020:1577-1586.
[7] ZHENG Z H,WANG P,LIU W,et al.Distance-IoU loss:faster and better learning for bounding box regression[C]//Proceedings of the AAAI Conference on Artificial Intelligence.Palo Alto:AAAI Press,2020:12993-13000.
[8] HU J,SHEN L,ALBANIE S,et al.Squeeze-and-Excitation networks[J].IEEE transactions on pattern analysis and machine intelligence,2020,42(8):2011-2023.
[9] 孔韦韦，雷阳，任聪，等.基于改进型CNN的多聚焦图像融合方法[J].郑州大学学报(理学版),2019,51(2):29-33.KONG W W,LEI Y,REN C,et al.Method for multi-focus image fusion based on improved convolutional neural network[J].Journal of Zhengzhou university (natural science edition),2019,51(2):29-33.
[10] 赵文清，周震东，翟永杰.基于反卷积和特征融合的SSD小目标检测算法[J].智能系统学报，2020,15(2):310-316.ZHAO W Q,ZHOU Z D,ZHAI Y J.SSD small target detection algorithm based on deconvolution and feature fusion[J].CAAI transactions on intelligent systems,2020,15(2):310-316.
基本信息:
DOI：10.13705/j.issn.1671-6841.2021287
中图分类号:TP391.41;V351
引用信息:
[1]王阳,袁国武,瞿睿,等.基于改进YOLOv3的机场停机坪目标检测方法[J],2022,54(05):22-28.DOI:10.13705/j.issn.1671-6841.2021287.
基金信息:
国家自然科学基金项目(11663007);; 云南省科技厅材料基因工程重大专项(2019ZE001-1,202002AB08001-6);; 云南省应用基础研究计划重点项目(202001BB050032)