基于NAS-Res的局部遮挡荷斯坦奶牛个体识别

姚冲; 李前; 刘刚; 吕树盛; 侯冲; 张淼

基于NAS-Res的局部遮挡荷斯坦奶牛个体识别

姚冲¹,
李前¹,
刘刚^1,2,
吕树盛³,
侯冲³,
张淼^1,2,

1. 中国农业大学智慧农业系统集成研究教育部重点实验室;
2. 中国农业大学农业农村部农业信息获取技术重点实验室;
3. 中国农业机械化科学研究院集团有限公司行业技术服务中心

基金项目:

国家自然科学基金项目(32272930)

详细信息

作者简介:
姚冲(1997—),女,博士生,主要从事计算机视觉技术研究,E-mail:yaochong@cau.edu.cn

通讯作者:
张淼(1982—),女,副教授,博士,主要从事农业信息获取技术研究,E-mail:zhangmiao@cau.edu.cn

中图分类号: S823;TP391.41
计量
- 文章访问数: 0
- HTML全文浏览量: 0
- PDF下载量: 0
出版历程
- 收稿日期: 2023-06-19
- 刊出日期: 2023-11-17

Individual Identification of Partially Occluded Holstein Cows Based on NAS-Res

1. Key Laboratory of Smart Agriculture Systems Integration, Ministry of Education, China Agricultural University;
2. Key Laboratory of Agricultural Information Acquisition Technology, Ministry of Agriculture and Rural Affairs,China Agricultural University;
3. Machinery Industry Technology Service Center, Chinese Academy of Agricultural Mechanization Sciences Group Co., Ltd.

摘要

摘要: 针对荷斯坦奶牛个体识别神经网络的人工调参成本高、泛化性差、效率低，难以实现局部遮挡条件下精准识别等问题，提出了一种基于ResNet框架和神经网络架构搜索(NAS)的自适应网络参数优化算法(NAS-Res)。首先，通过设计包含CBR＿K1、CBR＿K3、CBR＿K5和SkipConnect的操作集，配合密集连接路径，构成超参数网络。然后基于梯度下降的搜索策略，在多目标优化复合损失函数的约束下，强化了对低成本模型的设计。结果表明，NAS-Res在GPU上仅耗时6.18 h获得最佳架构，在包含168头奶牛局部遮挡侧面图像的PO-Cows数据集上，闭集验证准确率为90.18%,与ResNet-18、ResNet-34和ResNet-50相比提高5.04、3.02、14.92个百分点，而参数量分别降低5.9×10～5、1.069×10～7和1.317×10～7。在包含174头奶牛背部图像的Cows2021数据集上闭集验证准确率为99.25%。此外，NAS-Res可忽略PO-Cows数据集规模变化的影响，牛只数量在50～168头之间变化时，Top-1准确率和Top-5准确率变化幅度仅为1.51、1.01个百分点，适用性较强。总体而言，NAS-Res算法实现了对局部遮挡奶牛的精准个体识别，本研究可为复杂背景下畜禽个体识别提供技术参考。
- 荷斯坦奶牛 /
- 局部遮挡 /
- 神经网络架构搜索 /
- 卷积神经网络 /
- 个体识别
Abstract: The Holstein cow individual recognition network has the problems of high parameter adjustment cost, poor generalization and low efficiency, and it is difficult to achieve accurate recognition under partial occlusion conditions.An adaptive network parameter optimization identification algorithm（NAS-Res） was proposed based on ResNet framework and neural network architecture search（NAS）. Firstly, a hyperparameter network was constructed by designing an operation set, including CBR＿K1, CBR＿K3, CBR＿K5, and SkipConnect, together with dense connection paths. Then the search strategy based on gradient descent strengthened the design of a low-cost model under the constraint of multi-objective optimization composite loss function. The results showed that NAS-Res only took 6.18 GPU hours to obtain the best architecture.On the PO-Cows dataset, which contained side images of 168 cows, NAS-Res achieved 90.18% Top-1 Acc. Compared with ResNet-18, ResNet-34, and ResNet-50, the accuracy was improved by 5.04 percentage points, 3.02 percentage points, and 14.92 percentage points, respectively, while the parameters were reduced by 5.9×10~5, 1.069×10~7, and 1.317×10~7, respectively.It achieved 99.25% accuracy on the Cows2021 dataset, which contained 174 back images of cows. In addition, NAS-Res can ignore the influence of the scale change of the PO-Cows dataset, and when the number of cattle was changed between 50 and 168, the change range of Top-1 Acc and Top-5 Acc was only 1.51 percentage points and 1.01 percentage points, which showed strong applicability. In general, the NAS-Res algorithm achieved accurate individual identification of partially occluded cows, and the research result can provide technical reference for individual identification of livestock and poultry under complex background.
- Holstein cows /
- partially occluded /
- neural architecture search /
- convolutional neural network /
- individual recognition

HTML全文

参考文献(34)

[1]	黄小平,冯涛,郭阳阳,等.基于改进YOLO v5s的轻量级奶牛体况评分方法[J].农业机械学报,2023,54(6):287-296.HUANG Xiaoping,FENG Tao,GUO Yangyang,et al.Lightweight dairy cow body condition scoring method based on improved YOLO v5s[J].Transactions of the Chinese Society for Agricultural Machinery,2023,54(6):287-296.(in Chinese)
[2]	FERREIRA R E P,BRESOLIN T,ROSA G J M,et al.Using dorsal surface for individual identification of dairy calves through 3D deep learning algorithms[J].Computers and Electronics in Agriculture,2022,201:107272.
[3]	何东健,刘建敏,熊虹婷,等.基于改进YOLO v3模型的挤奶奶牛个体识别方法[J].农业机械学报,2020,51(4):250-260.HE Dongjian,LIU Jianmin,XIONG Hongting,et al.Individual identification of dairy cows based on improved YOLO v3[J].Transactions of the Chinese Society for Agricultural Machinery,2020,51(4):250-260.(in Chinese)
[4]	ZHENG Z,LI J,QIN L.YOLO-BYTE:an efficient multi-object tracking algorithm for automatic monitoring of dairy cows[J].Computers and Electronics in Agriculture,2023,209:107857.
[5]	KUMAR S,SINGH S K.Cattle recognition:a new frontier in visual animal biometrics research[J].Proceedings of the National Academy of Sciences,India section A:Physical Sciences,2020,90(4):689-708.
[6]	WILLIAMS L R,FOX D R,BISHOP-HURLEY G J,et al.Use of radio frequency identification (RFID) technology to record grazing beef cattle water point use[J].Computers and Electronics in Agriculture,2019,156:193-202.
[7]	高云.基于无线传感器网络的猪运动行为监测系统研究[D].武汉:华中农业大学,2014.GAO Yun.Study of pig behavior monitoring system based on WSN[D].Wuhan:Huazhong Agricultural University,2014.(in Chinese)
[8]	蔡骋,宋肖肖,何进荣.基于计算机视觉的牛脸轮廓提取算法及实现[J].农业工程学报,2017,33(11):171-177.CAI Cheng,SONG Xiaoxiao,HE Jinrong.Algorithm and realization for cattle face contour extraction based on computer vision[J].Transactions of the CSAE,2017,33(11):171-177.(in Chinese)
[9]	杨蜀秦,刘杨启航,王振,等.基于融合坐标信息的改进YOLO V4模型识别奶牛面部[J].农业工程学报,2021,37(15):129-135.YANG Shuqin,LIU Yangqihang,WANG Zhen,et al.Improved YOLO V4 model for face recognition of diary cow by fusing coordinate information[J].Transactions of the CSAE,2021,37(15):129-35.(in Chinese)
[10]	HU H,DAI B,SHEN W,et al.Cow identification based on fusion of deep parts features[J].Biosystems Engineering,2020,192:245-256.
[11]	SUN S,YANG S,ZHAO L.Noncooperative bovine iris recognition via SIFT[J].Neurocomputing,2013,120:310-317.
[12]	KUMAR S,SINGH S K,SINGH R S,et al.Real-time recognition of cattle using animal biometrics[J].Journal of Real-Time Image Processing,2017,13:505-526.
[13]	司永胜,王朝阳,张艳,等.基于深度与传统特征融合的非限制条件下奶牛个体识别[J].农业机械学报,2023,54(6):272-279.SI Yongsheng,WANG Zhaoyang,ZHANG Yan,et al.Individual identification of dairy cows under unrestricted conditions based on fusion of deep and traditional features[J].Transactions of the Chinese Society for Agricultural Machinery,2023,54(6):272-279.(in Chinese)
[14]	SHEN W,HU H,DAI B,et al.Individual identification of dairy cows based on convolutional neural networks[J].Multimedia Tools and Applications,2020,79:14711-14724.
[15]	KRIZHEVSKY A,SUTSKEVER I,HINTON G E.ImageNet classification with deep convolutional neural networks[J].Communications of the ACM,2017,60(6):84-90.
[16]	DU Y,KOU Y,LI B,et al.Individual identification of dairy cows based on deep learning and feature fusion[J].Animal Science Journal,2022,93(1):e13789.
[17]	SIMONYAN K,ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[J].arXiv preprint arXiv:1409.1556,2014.
[18]	LI Z,GE C,SHEN S,et al.Cow individual identification based on convolutional neural network[C]// Proceedings of the 2018 International Conference on Algorithms,Computing and Artificial Intelligence.Association for Computing Machinery,2018:10.1145/3302425.3302460.
[19]	TABAK M A,NOROUZZADEH M S,WOLFSON D W,et al.Improving the accessibility and transferability of machine learning algorithms for identification of animals in camera trap images:MLWIC2[J].Ecology and Evolution,2020,10(19):10374-10383.
[20]	HE K,ZHANG X,REN S,et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016.
[21]	WEBER F D L,WEBER V A D M,MENEZES G V,et al.Recognition of Pantaneira cattle breed using computer vision and convolutional neural networks[J].Computers and Electronics in Agriculture,2020,175:105548.
[22]	HUANG G,LIU Z,VAN DER MAATEN L,et al.Densely connected convolutional networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2017.
[23]	SZEGEDY C,IOFFE S,VANHOUCKE V,et al.Inception-v4,inception-resnet and the impact of residual connections on learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence,2017.
[24]	MAHMUD M S,ZAHID A,DAS A K,et al.A systematic literature review on deep learning applications for precision cattle farming[J].Computers and Electronics in Agriculture,2021,187:106313.
[25]	彭阳翔,杨振标,闫奎友,等.从人工到智能:牛个体识别技术研究进展[J].中国畜牧兽医,2023,50(5):1855-1866.PENG Yangxiang,YANG Zhenbiao,YAN Kuiyou,et al.From artificial to intelligent:research progress of individual idendification technology for cattle[J].China Animal Husbandry & Veterinary Medicine,2023,50(5):1855-1866.(in Chinese)
[26]	邓渭铭,杨铁军,李纯纯,等.基于神经网络架构搜索的铭牌目标检测方法[J].图学学报,2023,44(4):718-727.DENG Weiming,YANG Tiejun,LI Chunchun,et al.Object detection for nameplate based on neural architecture search[J].Journal of Graphics,2023,44(4):718-727.(in Chinese)
[27]	REN P,XIAO Y,CHANG X,et al.A comprehensive survey of neural architecture search:challenges and solutions[J].ACM Computing Surveys (CSUR),2021,54(4):1-34.
[28]	LIU H,SIMONYAN K,YANG Y.Darts:differentiable architecture search[J].arXiv preprint arXiv:1806.09055,2018.
[29]	ZOPH B,VASUDEVAN V,SHLENS J,et al.Learning transferable architectures for scalable image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2018.
[30]	BARBEDO J G A,KOENIGKAN L V,SANTOS T T,et al.A study on the detection of cattle in UAV images using deep learning[J].Sensors,2019,19(24):5436.
[31]	BARBEDO J G A,KOENIGKAN L V,SANTOS P M,et al.Counting cattle in UAV images—dealing with clustered animals and animal/background contrast changes[J].Sensors,2020,20(7):2126.
[32]	LIU C,ZOPH B,NEUMANN M,et al.Progressive neural architecture search[C]//Proceedings of the European Conference on Computer Vision (ECCV),2018.
[33]	YAO L,HU Z,LIU C,et al.Cow face detection and recognition based on automatic feature extraction algorithm[C]//Proceedings of the ACM Turing Celebration Conference.Association for Computing Machinery,2019:95.10.1145/3321408.3322628.
[34]	FANG J,SUN Y,ZHANG Q,et al.Densely connected search space for more flexible neural architecture search[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020.