Journal of Computer Applications ›› 2021, Vol. 41 ›› Issue (5): 1332-1336.DOI: 10.11772/j.issn.1001-9081.2020071126

Special Issue: 人工智能

• Artificial intelligence • Previous Articles     Next Articles

Yak face recognition algorithm of parallel convolutional neural network based on transfer learning

CHEN Zhengtao1, HUANG Can1, YANG Bo2, ZHAO Li3, LIAO Yong4   

  1. 1. Sichuan State-owned Assets Investment and Management Company Limited, Chengdu Sichuan 610031, China;
    2. Sichuan SDIC Modern Agriculture and Animal Husbandry Industry Company Limited, Chengdu Sichuan 610041, China;
    3. Chengdu Simu-Tech Science and Technology Development Company Limited, Chengdu Sichuan 610041, China;
    4. School of Microelectronics and Communication Engineering, Chongqing University, Chongqing 400044, China
  • Received:2020-07-30 Revised:2020-09-14 Online:2021-05-10 Published:2020-10-19
  • Supported by:
    This work is partially supported by Scientific Research Project of Sichuan Plateau Ecological Industry Development Research Center in 2020 (YJZX-2020-2).


陈争涛1, 黄灿1, 杨波2, 赵立3, 廖勇4   

  1. 1. 四川省国有资产投资管理有限责任公司, 成都 610031;
    2. 四川省国投现代农牧业产业有限公司, 成都 610041;
    3. 成都希盟泰克科技发展有限公司, 成都 610041;
    4. 重庆大学 微电子与通信工程学院, 重庆 400044
  • 通讯作者: 廖勇
  • 作者简介:陈争涛(1968-),男,重庆人,信息分析师,主要研究方向:农业经济;黄灿(1990-),男,湖北天门人,经济师,硕士,主要研究方向:产业投资与运行;杨波(1975-),男,重庆人,工程师,硕士,主要研究方向:通信项目管理;赵立(1968-),男,重庆人,高级工程师,硕士,主要研究方向:虚拟现实;廖勇(1982-),男,四川自贡人,副研究员,博士,CCF杰出会员,主要研究方向:移动通信、人工智能。
  • 基金资助:

Abstract: In order to realize accurate management of yaks during the process of yak breeding, it is necessary to recognize the identities of the yaks. Yak face recognition is a feasible method of yak identification. However, the existing yak face recognition algorithms based on neural networks have the problems such as too many features in the yak face dataset and long training time of neural networks. Therefore, based on the method of transfer learning and combined with the Visual Geometry Group (VGG) network and Convolutional Neural Network (CNN), a Parallel CNN (Parallel-CNN) algorithm was proposed to identify the facial information of yaks. Firstly, the existing VGG16 network was used to perform transfer learning to the yak face image data and extract the yaks' facial information features for the first time. Then, the dimensional transformation was performed to the extracted features at different levels, and the processed features were inputted into the parallel-CNN for the secondary feature extraction. Finally, two separated fully connected layers were used to classify the yak face images. Experimental results showed that Parallel-CNN was able to recognize yak faces with different angles, illuminations and poses. On the test dataset with 90 000 yak face images of 300 yaks, the recognition accuracy of the proposed algorithm reached 91.2%. The proposed algorithm can accurately recognize the identities of the yaks, and can help the yak farm to realize the intelligent management of the yaks.

Key words: yak face recognition, deep learning, transfer learning, Convolutional Neural Network (CNN), parallel network

摘要: 为了在牦牛养殖过程中对牦牛实现精确管理,需要对牦牛的身份进行识别,而牦牛脸识别是一种可行的牦牛身份识别方式。然而已有的基于神经网络的牦牛脸识别算法中存在牦牛脸数据集特征多、神经网络训练时间长的问题,因此,借鉴迁移学习的方法并结合视觉几何组网络(VGG)和卷积神经网络(CNN),提出了一种并行CNN(Parallel-CNN)算法用来识别牦牛的面部信息。首先,利用已有的VGG16网络对牦牛脸图像数据进行迁移学习以及初次提取牦牛的面部信息特征;然后,将提取到的不同层次的特征进行维度变换并输入到Parallel-CNN中进行二次特征提取;最后,利用两个分离的全连接层对牦牛脸图像进行分类。实验结果表明:Parallel-CNN能够对不同角度、光照和姿态的牦牛脸进行识别,在含有300头牦牛的90 000张牦牛脸图像的测试数据集上,所提算法的识别准确率达到91.2%。所提算法可以对牦牛身份进行精确识别,从而帮助牦牛养殖场实现对牦牛的智能化管理。

关键词: 牦牛脸识别, 深度学习, 迁移学习, 卷积神经网络, 并行网络

CLC Number: