郭祥云,刘雨欣,孔晓亚,冯嘉朔

• 1:北京信息科技大学管理科学与工程学院

摘要(Abstract)：

家猫的行为模式可直接反映其生理和心理健康状态，准确识别家猫日常行为有助于及时发现宠物异常、提升宠物福利水平。现有动物行为识别方法多聚焦于畜牧动物，难以有效应对家猫行为动作幅度小、时序变化快、背景环境复杂的挑战。针对上述问题，提出了SMTSlowFast模型，该模型基于SlowFast框架，在Slow路径与Fast路径中协同引入时序移位模块（temporal shift module, TSM）与SEAM注意力模块（SE-SimAM attention module）。TSM通过沿时间维度对部分通道特征进行前后移位，实现相邻帧信息的高效交互，增强对行为动态时序变化的建模能力；SEAM融合通道级注意力与神经元级重要性的三维权重，突出关键行为区域并有效抑制复杂背景干扰。构建了多场景家猫行为数据集（multi-scene domestic cat behavior dataset, MDCBD）对SMT-SlowFast模型进行验证。实验结果表明：针对8类行为，SMT-SlowFast模型准确率达到94.60%，优于多个基准模型，且对理毛、奔跑和行走等行为的识别性能提升尤为显著。该研究为非接触式家猫行为监测提供了高效、鲁棒的技术路径。

关键词(KeyWords)： 家猫;行为识别;SlowFast;时空特征;计算机视觉

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 郭祥云,刘雨欣,孔晓亚,冯嘉朔

DOI: 10.16508/j.cnki.11-5866/n.2026.04.008

参考文献(References)：

• [1]派读宠物行业大数据平台. 2026年中国宠物行业白皮书（消费报告）[R].上海：派读宠物行业大数据平台，2025.Paidu pet industry big data platform. 2026 white paper on China's pet industry(consumption report)[R]. Shanghai:Paidu pet industry big data platform, 2025.(in Chinese)
• [2]Foreman-Worsley R,Finka L R,Ward S J,et al. Indoors or outdoors? An international exploration of owner demographics and decision making associated with lifestyle of pet cats[J].Animals,2021,11(2):253.
• [3]Barwick J,Lamb D,Dobos R,et al. Predicting lameness in sheep activity using tri-axial acceleration signals[J]. Animals,2018,8(1):12.
• [4]Chen Guanyu,Takegawa Y,Matsumura K,et al. Cat and dog behavior recognition method using deep learning approach based on inertial measurement unit sensor data[J]. Sensors&Materials,2025,37(3):1073-1098.
• [5]Rahman A,Smith D V,Little B,et al. Cattle behaviour classification from collar,halter,and ear tag sensors[J].Information Processing in Agriculture,2018,5(1):124-133.
• [6]Chen Chen,Zhu Weixing,Norton T. Behaviour recognition of pigs and cattle:journey from computer vision to deep learning[J]. Computers and Electronics in Agriculture,2021,187:106255.
• [7]杨断利，齐俊林，陈辉，等.基于改进YOLO v8n模型的散养蛋鸡个体行为识别方法与差异分析[J].农业机械学报，2024,55(11):112-123.Yang Duanli,Qi Junlin,Chen Hui,et al. Individual behavioral identification and differential analysis of free-range laying hens based on improved YOLO v8n model[J]. Transactions of the Chinese Society for Agricultural Machinery,2024,55(11):112-123.(in Chinese)
• [8]王亚彬，徐爱俊，周素茵，等.基于Byte的生猪多目标跟踪算法[J].农业工程学报，2025,41(7):145-155.Wang Yabin,Xu Aijun,Zhou Suyin,et al. Multi-object tracking of pig behavior using Byte algorithm[J]. Transactions of the Chinese Society of Agricultural Engineering,2025,41(7):145-155.(in Chinese)
• [9]Long Ting,Yang Rongchuan,You Xu,et al. FSCA-YOLO:an enhanced yolo-based model for multi-target dairy cow behavior recognition[J]. Animals,2024,14(17):2631.
• [10]Lauer J, Zhou Mu, Ye Shaokai, et al. Multi-animal pose estimation, identification and tracking with DeepLabCut[J].Nature Methods,2022,19(4):496-504.
• [11]Fang Cheng,Zhang Tiemin,Zheng Haikun,et al. Pose estimation and behavior classification of broiler chickens based on deep neural networks[J]. Computers and Electronics in Agriculture,2020,180:105863.
• [12]Feng Liqi,Zhao Yaqin,Sun Yichao,et al. Action recognition using a spatial-temporal network for wild felines[J]. Animals,2021,11(2):485.
• [13]Ma Shuaifei, Zhang Qianru,Li Tengfei,et al. Basic motion behavior recognition of single dairy cow based on improved Rexnet 3D network[J]. Computers and Electronics in Agriculture,2022,194:106772.
• [14]Yu Wenbo,Chen Yongqi,Wang Xiang,et al. YoloSortC3D:a YOLOv8, DeepSort, and C3D-based method for sheep rumination behavior recognition[J]. Applied Sciences,2024,14(23):10801.
• [15]Wang Jinwei,Du Zhiguo,Wen Bin,et al. EML-SlowFast:a behavior recognition model for lion-head goose[J]. Poultry Science,2025,104(8):105221.
• [16]Sun Gang,Liu Tonghai,Zhang Hang,et al. Basic behavior recognition of yaks based on improved SlowFast network[J].Ecological Informatics,2023,78:102313.
• [17]Li Ran,Dai Baisheng,Hu Yuhang,et al. Multi-behavior detection of group-housed pigs based on YOLOX and SCTSSlowFast[J]. Computers and Electronics in Agriculture,2024,225:109286.
• [18]Zhang Junpeng,Bai Zihan,Wei Yifan,et al. Behavior detection of dairy goat based on YOLO11 and ELSlowFast-LSTM[J].Computers and Electronics in Agriculture,2025,234:110224.
• [19]Liu Yanhong,Zhou Fang,Zheng Wenxin,et al. Sleeping and eating behavior recognition of horses based on an improved SlowFast network[J]. Sensors,2024,24(23):7791.
• [20]Zhang Bo,Li Jia,Kang Feilong,et al. Detecting dairy cattle protective behaviors via a multi-stage attention SlowFast network[J]. Animals,2026,16(9):1321.
• [21]Lee J H,Kang B M. Real-time cat behavior recognition system using two-stream YOLO[C]//Annual Conference of Korea Information Processing Society,2019:574-576.
• [22]Sui Zhuoneng,Hong Haoran,Shimtoku D,et al. CatAction:deep learning for enhancing emotional cat-human interactions through the posture-based determination of the degrees of kittens’ defensive and offensive aggressions[C]//Proceedings of the International Conference on Animal-Computer Interaction.New York:Association for Computing Machinery,2024:1-9.
• [23]Stanton L A,Sullivan M S,Fazio J M. A standardized ethogram for the felidae:a tool for behavioral researchers[J]. Applied Animal Behaviour Science,2015,173:3-16.
• [24]Feichtenhofer C, Fan Haoqi, Malik J, et al. SlowFast networks for video recognition[C]//2019 IEEE/CVF International Conference on Computer Vision(ICCV),2019:6201-6210.
• [25]Hara K,Kataoka H,Satoh Y. Can spatiotemporal 3D CNNs retrace the history of 2D CNNs and imagenet?[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition,2018:6546-6555.
• [26]Hu Jie,Shen Li,Albanie S,et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(8):2011-2023.
• [27]Yang Lingxiao,Zhang Ruyuan,Li Lida,et al. SimAM:a simple,parameter-free attention module for convolutional neural networks[C]//Proceedings of the 38th International Conference on Machine Learning. New York:PMLR,2021:11863-11874.
• [28]Lin Ji, Gan Chuang, Wang Kuan, et al. TSM:temporal shift module for efficient and scalable video understanding on edge devices[PP/OL].(2021-09-27)[2025-02-24]. https：//arxiv.org/abs/2109. 13227.
• [29]Selvaraju R R,Cogswell M,Das A,et al. Grad-CAM:visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision,2020,128(2):336-359.
• [30]Wang Limin,Xiong Yuanjun,Wang Zhe,et al. Temporal segment networks:towards good practices for deep action recognition[C]//Computer Vision-ECCV 2016,2016:20-36.
• [31]Simonyan K,Zisserman A. Two-stream convolutional networks for action recognition in videos[C]//Proceedings of the 28th International Conference on Neural Information Processing Systems. Cambridge:MIT Press,2014:568-576.
• [32]Carreira J,Zisserman A. Quo vadis, action recognition? A new model and the Kinetics dataset[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR),2017:4724-4733.
• [33]Tran D,Wang Heng,Torresani L,et al. A closer look at spatiotemporal convolutions for action recognition[C]//2018IEEE/CVF Conference on Computer Vision and Pattern Recognition,2018:6450-6459.