基于改进的YOLOv3和Facenet的无人机影像人脸识别

doi:10.7523/j.ucas.2021.0019

中国科学院大学学报 ›› 2023, Vol. 40 ›› Issue (1): 93-100.DOI: 10.7523/j.ucas.2021.0019

• 电子信息与计算机科学 • 上一篇

基于改进的YOLOv3和Facenet的无人机影像人脸识别

高锦风^1,2, 陈玉¹, 魏永明¹, 李剑南^1,2

1. 中国科学院空天信息创新研究院, 北京 100094;
2. 中国科学院大学, 北京 100049

收稿日期:2021-01-15 修回日期:2021-03-12 发布日期:2021-05-31
通讯作者: 魏永明,E-mail:weiym@aircas.ac.cn
基金资助:
兵团科技攻关项目(2017DB005-01)和国家重点研发计划项目(2017YFC1500902)资助

Unmanned aerial vehicle image face recognition based on improved YOLOv3 and Facenet

GAO Jinfeng^1,2, CHEN Yu¹, WEI Yongming¹, LI Jiannan^1,2

1. Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100094, China;
2. University of Chinese Academy of Sciences, Beijing 100049, China

Received:2021-01-15 Revised:2021-03-12 Published:2021-05-31

摘要/Abstract

摘要： 基于无人机影像的高精度人脸识别在应急救援、嫌疑人员跟踪等场景中发挥着重要作用。深度学习卷积神经网络以其较高的精度和较少的人为干扰被广泛应用于目标检测识别领域，能很好地应用于无人机影像人脸识别任务中。探究在无人机嫌疑人员识别应用场景下利用卷积网络进行人脸高精度识别，用改进后的YOLOv3(you only look once)进行无人机影像的人脸检测，将得到的预测框对齐后输入到经典的Facenet人脸识别网络中进行目标身份的判定。实验对比了改进后的YOLOv3、原始YOLOv3和MTCNN(multi-task convolutional neural network)的检测效果以及结合Facenet进行人脸识别的效果。结果表明：1)改进后的YOLOv3相对于原始YOLOv3不仅精度和召回率得到提升，而且模型参数量有所减少，无人机影像的漏检和错检现象也轻于原始YOLOv3；此外，改进后的YOLOv3相对MTCNN的AP(average precision)提升9.49%，检测速度也约是MTCNN的3倍；2)改进后的YOLOv3+Facenet相对于原始YOLOv3+Facenet及MTCNN+Facenet对人脸的区分能力更强，精度更高，对遮挡以及模糊的鲁棒性也更强。

关键词: YOLOv3, Facenet, 人脸识别, 无人机

Abstract: High precision face recognition based on unmanned aerial vehicle (UAV) images plays an important role in emergency rescue, suspect tracking, and other scenes. Deep learning convolutional neural network is widely used in the field of target detection and recognition because of its high accuracy and less human interference, which can be well applied to UAV image face recognition tasks. This paper explores the use of convolution networks for high-precision face recognition in UAV application scenarios, uses the improved YOLOv3(you only look once) for face detection of UAV images, and inputs the prediction boxes into the classic Facenet network to determine the target identity. Through experiments, this paper compares the detection effect of the improved YOLOv3, the original YOLOv3, and the MTCNN (multi-task convolutional neural network), and also compares the face recognition effect of the three models combined with Facenet. The experimental results show that: 1) compared with the original YOLOv3, the improved YOLOv3 improves the accuracy and recall rate, reduces the number of model parameters; besides, the phenomenon of missing and wrong detection of the improved YOLOv3 for UAV image is less than that of the original YOLOv3; moreover, the AP (average precision) of improved YOLOv3 is 9.49% higher than that of MTCNN, and the detection speed is about 3 times of MTCNN; 2) compared with the original YOLOv3+Facenet and MTCNN+Facenet, the improved YOLOv3+Facenet has stronger ability to distinguish faces and higher accuracy, and has stronger robustness to occlusion and blur.

Key words: YOLOv3, Facenet, face recognition, unmanned aerial vehicle

中图分类号:

TP79

高锦风, 陈玉, 魏永明, 李剑南. 基于改进的YOLOv3和Facenet的无人机影像人脸识别[J]. 中国科学院大学学报, 2023, 40(1): 93-100.

GAO Jinfeng, CHEN Yu, WEI Yongming, LI Jiannan. Unmanned aerial vehicle image face recognition based on improved YOLOv3 and Facenet[J]. Journal of University of Chinese Academy of Sciences, 2023, 40(1): 93-100.

参考文献

[1] 李志远. 人脸识别技术研究现状综述[J]. 电子技术与软件工程, 2020, 27(13): 106-107.
[2] 邓良, 许庚林, 李梦杰, 等. 基于深度学习与多哈希相似度加权实现快速人脸识别[J]. 计算机科学, 2020, 47(9): 163-168.
[3] 李刚, 高政. 人脸自动识别方法综述[J]. 计算机应用研究, 2003, 20(8): 4-9,40.DOI：10.3969/j.issn.1001-3695.2003.08.002.
[4] 张翠平, 苏光大. 人脸识别技术综述[J]. 中国图象图形学报, 2000, 5(11): 885-894.DOI：10.3969/j.issn.1006-8961.2000.11.001.
[5] Girshick R. Fast R-CNN[C]//2015 IEEE International Conference on Computer Vision (ICCV). December 7-13, 2015, Santiago, Chile. IEEE, 2015: 1440-1448.DOI：10.1109/ICCV.2015.169.
[6] 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. IEEE, 2017,39(6): 1137-1149.DOI：10.1109/TPAMI.2016.2577031.
[7] He K M, Gkioxari G, Dollár P, et al. Mask R-CNN[C]//2017 IEEE International Conference on Computer Vision (ICCV). October 22-29, 2017, Venice, Italy. IEEE, 2017: 2980-2988.
[8] Redmon J, Divvala S, Girshick R, et al. You only look once: unified, real-time object detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 27-30, 2016, Las Vegas, NV, USA. IEEE, 2016: 779-788.DOI：10.1109/CVPR.2016.91.
[9] Redmon J, Farhadi A. YOLO9000: better, faster, stronger[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). July 21-26, 2017, Honolulu, HI, USA. IEEE, 2017: 6517-6525.DOI：10.1109/CVPR.2017.690.
[10] Redmon J, Farhadi A. YOLOv3: an incremental improvement [EB/OL]. 2018: arXiv: 1804.02767. (2018-04-08) [2021-03-08]. https://arxiv.org/abs/1804.02767.
[11] 高刘雅, 孙冬, 卢一相. 基于轻量级注意机制的人脸检测算法[J]. 激光与光电子学进展, 2021, 58(2): 130-138.DOI：10.3788/LOP202158.0210010.
[12] 潘浩然. 基于改进损失函数的YOLOV3的人脸检测[D]. 南昌: 南昌大学, 2020.
[13] Schroff F, Kalenichenko D, Philbin J. FaceNet: a unified embedding for face recognition and clustering[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 7-12, 2015, Boston, MA, USA. IEEE, 2015: 815-823.DOI：10.1109/CVPR.2015.7298682.
[14] Dai J F, He K M, Sun J. Instance-aware semantic segmentation via multi-task network cascades[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 27-30, 2016, Las Vegas, NV, USA. IEEE, 2016: 3150-3158.DOI：10.1109/CVPR.2016.343.
[15] 刘长伟. 基于MTCNN和Facenet的人脸识别[J].邮电设计技术, 2020, 63(2): 32-38.DOI：10.12045/j.issn.1007-3043.2020.02.008.
[16] 李林峰, 李春青, 田博源, 等. 基于MTCNN的FaceNet架构的人脸识别考勤系统设计与实现[J]. 电脑知识与技术, 2020, 16(27): 181-183.DOI：10.14004/j.cnki.ckt.2020.2926.
[17] Yang S, Luo P, Loy C C, et al. WIDER FACE: a face detection benchmark[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 27-30, 2016, Las Vegas, NV, USA. IEEE, 2016: 5525-5533.DOI：10.1109/CVPR.2016.596.
[18] Hu J, Shen L, Albanie S, et al. Squeeze-and-excitation networks[C]//IEEE Transactions on Pattern Analysis and Machine Intelligence. IEEE, 2020,42(8): 2011-2023.
[19] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). June 27-30, 2016, Las Vegas, NV, USA. IEEE, 2016: 770-778.
[20] Xie S N, Girshick R, Dollár P, et al. Aggregated residual transformations for deep neural networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). July 21-26, 2017, Honolulu, HI, USA. IEEE, 2017: 5987-5995.
[21] Zheng Z H, Wang P, Liu W, et al. Distance-IoU loss: faster and better learning for bounding box regression[EB/OL]. 2019,arXiv:1911.08287.(2019-11-19)[2021-03-10]. http://arxiv.org/abs/1911.08287.
[22] Rezatofighi H, Tsoi N, Gwak J, et al. Generalized intersection over union: a metric and a loss for bounding box regression[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). June 15-20, 2019, Long Beach, CA, USA. IEEE, 2019: 658-666.DOI：10.1109/CVPR.2019.00075.
[23] 谢梦, 刘伟, 杨梦圆, 等. 深度卷积神经网络支持下的遥感影像飞机检测[J]. 测绘通报, 2019, 65(6): 19-23.DOI：10.13474/j.cnki.11-2246.2019.0177.
[24] Zhang P, Su W H. Statistical inference on recall, precision and average precision under random selection[C]//2012 9th International Conference on Fuzzy Systems and Knowledge Discovery. May 29-31, 2012, Chongqing, China. IEEE, 2012: 1348-1352.DOI：10.1109/FSKD.2012.6234049.

基于改进的YOLOv3和Facenet的无人机影像人脸识别

Unmanned aerial vehicle image face recognition based on improved YOLOv3 and Facenet

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 6

编辑推荐

Metrics

本文评价

访问统计

联系我们

[1]	刘梦冰, 邱玲, 梁晓雯. 基于Matern硬核点过程的无人机辅助蜂窝网络吞吐量分析[J]. 中国科学院大学学报, 2022, 39(5): 704-711.
[2]	宋孝成, 刘晓培, 陆疌. 未知环境中无人机实时导航的人工势场方法[J]. 中国科学院大学学报, 2022, 39(3): 393-402.
[3]	顾育津, 宋孝成, 刘晓培, 陆疌. 基于拉普拉斯人工势场的无人机避障控制[J]. 中国科学院大学学报, 2020, 37(5): 681-687.
[4]	李志新, 王心源, 刘传胜, 骆磊. 海南东寨港海底村庄遗址遥感观测与分析[J]. 中国科学院大学学报, 2019, 36(6): 803-809.
[5]	马娇, 董勇伟, 李原, 李凌霄, 杨杰芳. 多旋翼无人机微多普勒特性分析与特征提取[J]. 中国科学院大学学报, 2019, 36(2): 235-243.
[6]	曹林王东峰邹谋炎. 基于隐马尔可夫模型的人像鉴别算法[J]. 中国科学院大学学报, 2006, 23(1): 52-59.