Skip to main content
ARS Home » Southeast Area » Mississippi State, Mississippi » Crop Science Research Laboratory » Genetics and Sustainable Agriculture Research » Research » Publications at this Location » Publication #380491

Research Project: Closing the Yield Gap of Cotton, Corn, and Soybean in the Humid Southeast with More Sustainable Cropping Systems

Location: Genetics and Sustainable Agriculture Research

Title: Practices and applications of convolutional neural network-based computer vision systems in animal farming: a review

Author
item LI, GUOMING - MISSISSIPPI STATE UNIVERSITY
item Huang, Yanbo
item CHEN, ZHIQIAN - MISSISSIPPI STATE UNIVERSITY
item CHESSER JR, GRAY - MISSISSIPPI STATE UNIVERSITY
item ZHAO, YANG - UNIVERSITY OF TENNESSEE
item Purswell, Joseph
item LINHOSS, JOHN - MISSISSIPPI STATE UNIVERSITY

Submitted to: Sensors
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/19/2021
Publication Date: 2/21/2021
Citation: Li, G., Huang, Y., Chen, Z., Chesser Jr, G.D., Zhao, Y., Purswell, J.L., Linhoss, J. 2021. Practices and applications of convolutional neural network-based computer vision systems in animal farming: a review. Sensors. 21(4):1492. https://doi.org/10.3390/s21041492.
DOI: https://doi.org/10.3390/s21041492

Interpretive Summary: Machine learning has been developed into deep learning for applications in agriculture. This paper is a collaborative investigation of scientists in Mississippi State University, USDA ARS Genetics and Sustainable Agriculture Research Unit, University of Tennessee and USDA ARS Poultry Research Unit for summarizing and exploring convolutional neural networks, the most popular deep learning neural networks, for computer vision of animal farming to improve animal management. Through the investigation various CNN-based computer vision systems in animal farming are described and discussed for their advantages and disadvantages with the tasks involved in image classification, object detection, semantic/instance segmentation, pose estimation, and tracking with the targets of cattle, sheep/goat, pig, and poultry. Also, the practices of system development are investigated on sensor settings, system configuration, model structure and data processing. At last, future research is discussed to develop and improve the systems in better animal farming. The research of this paper definitely provide useful, comprehensive information for CNN deep learning for computer vision of animal farming, and at the same time the information can be transferred to and shared with other agricultural research, for example, precision crop farming through deep learning.

Technical Abstract: Convolutional neural network (CNN)-based computer vision systems have been increasingly applied into animal farming to improve animal management, but current knowledge, practices, limitations, and solutions of the applications remain unclear. The objective of this study was to systematically investigate various CNN-based computer vision systems in animal farming. In deep learning computer vision numerous tasks were involved, including image classification, object detection, semantic/instance segmentation, pose estimation, and tracking. Cattle, sheep/goat, pig, and poultry were the major targets of farm animal species. In this research practices of preparations for system development, including camera settings, inclusion of variations for data recordings, choices of graphic processing units, image preprocessing, and data labelling were summarized. CNN architectures were organized based on the computer vision tasks mentioned above in animal farming. Involved strategies of algorithm development included distribution of development data, data augmentation, hyperparameter tuning, selection of evaluation metrics, and judgment of model performance. Besides practices in optimizing CNN-based computer vision systems, system applications were also organized based on year, country, animal species, and purposes. Finally, several future researches were provided for developing and improving CNN-based computer vision systems in animal farming, so that the developed systems can better assist in precision animal management and improve efficiency of animal productivity.