数字农科院2.0

Lightweight model for beef cattle behavior recognition from quadruped robot video in grassland pastures

文献类型： 外文期刊

作者： Wei, Peigang;Sun, Wei;Cao, Shanshan;Kong, Fantao

作者机构：

关键词： Beef cattle;Behavior recognition;Grassland pastures;Quadruped robot vision;Edge computing

期刊名称： COMPUTERS AND ELECTRONICS IN AGRICULTURE

ISSN： 0168-1699

年卷期： 2025 年 242 卷

页码：

收录情况： SCIE（2025版） ; ; EI（2025版）

摘要： Accurate and rapid identification of typical cattle behaviors is fundamental to disease diagnosis, estrus monitoring, calving prediction, and health assessment. Existing machine vision approaches for behavior recognition in medium-to-large livestock (such as pigs, cattle, and sheep) are mainly tailored for indoor barn conditions. These methods perform poorly in outdoor grazing environments, where variable lighting, complex backgrounds, group clustering with occlusion, and motion blur pose substantial challenges. Our study proposes MASM-YOLO, a lightweight beef cattle behavior recognition model based on quadrupedal robots and edge computing. Using YOLOv11s as the baseline, we designed the Multi-Scale Focus and Extraction Network (MSFEN) to mitigate detection challenges caused by spatial scale differences and motion blur while enhancing cross-scale feature interaction. We further constructed the Adaptive Decomposition and Alignment Head (ADAH) to improve recognition accuracy in scenarios with clustering and occlusion. The lightweight feature extraction network StarNet optimizes the backbone structure, significantly reducing parameter count and computational load. The Inner-MPDIoU loss function is introduced to enhance the convergence and robustness of bounding box regression. Results demonstrate that MASM-YOLO achieves an mAP@0.5 of 90.4 % on the beef cattle behavior test set, surpassing the baseline model by 1.9 percentage points and significantly outperforming mainstream CNN and Transformer models. The model contains 4.0 M parameters and requires 18.2G FLOPs, representing reductions of 57.4 % and 14.6 % compared to the baseline. After TensorRT optimization on the NVIDIA Jetson Orin NX edge platform, MASM-YOLO achieves real-time inference at 36FPS while maintaining an mAP@0.5 of 89.6 %, validating its efficiency and feasibility on robotic platforms. This study demonstrates lightweight intelligent recognition of beef cattle behavior in grassland pastures through quadruped robot vision. Its uniqueness lies in constructing a new dataset and completing practical deployment verification on a robotic platform, providing valuable references and insights for automatic real-time livestock behavior recognition using ground-based mobile intelligent equipment in open grazing environments.

分类号：

  • 相关文献

[1]Cattle-ES3D: A spatiotemporal feature fusion method for detecting tachypnea and salivation behaviors in beef cattle. Fuyang Tian,Liyin Zhang,Ji Zhang,Shuaiyang Zhang,Shakeel Ahmed Soomro,Benhai Xiong,Weizheng Shen,Zhanhua Song,Yinfa Yan,Zhenwei Yu. 2025

[2]Research on Video Behavior Detection and Analysis Model for Sow Estrus Cycle Based on Deep Learning. Kaidong Lei,Bugao Li,Shan Zhong,Hua Yang,Hao Wang,Xiangfang Tang,Benhai Xiong. 2025

[3]Recognition of Foal Nursing Behavior Based on an Improved RT-DETR Model. Yanhong Liu,Fang Zhou,Wenxin Zheng,Tao Bai,Xinwen Chen,Leifeng Guo. 2025

[4]Intelligent recognition of basic camel behaviors based on DAF-Net for free-grazing camels. Zhou, Fang,Xian, Yuting,Guo, Leifeng,Peng, Cheng,Liu, Yanhong,Liu, Xiang,Xiao, Jing,Tian, Min. 2025

[5]Study of convolutional neural network-based semantic segmentation methods on edge intelligence devices for field agricultural robot navigation line extraction. Jiya Yu,Jiye Zhang,Aijing Shu,Yujie Chen,Jianneng Chen,Yongjie Yang,Wei Tang,Yanchao Zhang. 2023

[6]Speeding at the Edge: An Efficient and Secure Redactable Blockchain for IoT-Based Smart Grid Systems. Youshui Lu,Xiaojun Tang,Lei Liu,F. Richard Yu,Schahram Dustdar. 2023

[7]Real-Time Object Detection Based on UAV Remote Sensing: A Systematic Literature Review. Zhen Cao,Lammert Kooistra,Wensheng Wang,Leifeng Guo,João Valente. 2023

[8]GVC-YOLO: A Lightweight Real-Time Detection Method for Cotton Aphid-Damaged Leaves Based on Edge Computing. Zhenyu Zhang,Yunfan Yang,Xin Xu,Liangliang Liu,Jibo Yue,Ruifeng Ding,Yanhui Lu,Jie Liu,Hongbo Qiao. 2024

[9]A self-adaptive parallel image stitching algorithm for unmanned aerial vehicles in edge computing environments. Xin Xu,Li Zhang,Jibo Yue,Heming Zhong,Ying Wang,Jie Liu,Yanhui Lu,Hongbo Qiao. 2024

[10]QseedNet: A lightweight deep-learning framework for detection of quarantine weed seeds in imported soybean inspection. Jianwei Qiang,Xi Qiao,Yun Yu,Ze Liu,Zhihui Tian,Shuo Zhang. 2025

[11]Effect of wheat distillers dried grains with solubles (DDGS) as a replacement for barley grain and barley silage on ruminal pH and fermentation in finishing beef cattle. Li, Y. L.,Yang, W. Z.,He, M. L.,McAllister, T. A.,Beauchemin, K. A.. 2010

[12]Genetic Polymorphisms of Mc4R and IGF2 Gene Association with Feed Conversion Efficiency Traits in Beef Cattle. Du, Xin-hua,Chen, Cui,Yuan, Zheng-rong,Zhang, Li-min,Chen, Xiao-jie,Wang, Yan-hui,Gao, Xue,Zhang, Lu-pei,Gao, Hui-jiang,Li, Jun-ya,Xu, Shang-zhong. 2013

[13]Wheat distillers grains in feedlot cattle diets: Feeding behavior, growth performance, carcass characteristics, and blood metabolites. Yang, W. Z.,Li, Y. L.,McAllister, T. A.,Beauchemin, K. A.,Li, Y. L.,McKinnon, J. J.. 2012

[14]Association of genetic variations in the ACLY gene with growth traits in Chinese beef cattle. Li, M. N.,Guo, X.,Bao, P. J.,Wu, X. Y.,Ding, X. Z.,Chu, M.,Liang, C. N.,Yan, P.. 2016

[15]Association between PON1 gene SNPs and growth and carcass traits in beef cattle. Ji, A. C.,Huai, Y. H.,Zhou, Z. K.,Li, J. Y.,Zhang, L. P.,Xu, S. Z.,Gao, X.,Ren, H. Y.,Chen, J. B.,Ji, A. C.,Huai, Y. H.. 2008

[16]Effects of monensin, virginiamycin and sodium bicarbonate on rumen fermentation of beef cattle fed medium concentrate.. Wei, H. Y.,Wang, J. Q.,Li, C. H.,Bu, D. P.,Zhou, L. Y.. 2007

[17]Modified Bfat-1 gene and its biological verification in mice by hydrodynamic tail vein injection. Gao Xue,Du Xin-hua,Zhang Lu-pei,Chen Yan,Gao Hui-jiang,Xu Shang-zhong,Li Jun-ya,Du Xin-hua. 2016

[18]Association analysis of thyroglobulin gene variants with carcass and meat quality traits in beef cattle. Xu, Shang-Zhong,Gan, Qian-Fu,Hou, Guan-Yu,Xu, Shang-Zhong. 2008

[19]Bovine lipid metabolism related gene GPAM: Molecular characterization, function identification, and association analysis with fat deposition traits. Yu, Haibin,Zhao, Zhihui,Lu, Chuanyan,Yang, Runjun,Yu, Xianzhong,Li, Junya.

[20]Does the in situ ruminal degradation of feeds vary with the finishing ration fed to beef cattle?. Li, Y. L.,Yang, W. Z.. 2010

作者其他论文 更多>>