母羊舔羔与新生羔羊站立姿态的识别及关联分析

doi:10.16431/j.cnki.1671-7236.2025.08.024

摘要/Abstract

摘要： 【目的】试验旨在研究母羊舔羔和羔羊站立姿态智能检测、羔羊站立过程特征以及舔羔对羔羊站立的影响。【方法】收集30只母羊舔羔和羔羊站立过程视频，进行YOLOv5s模型训练。选取20只经产单胎萨能奶山羊及其子代新生羔羊，其中舔羔时长不足120 s的为对照组(Con组)，舔羔时间在120 s以上的为舔羔组(Lick组)，每组10只。基于训练好的YOLOv5s算法对母羊舔羔和羔羊站立姿态进行识别，依靠Farneback光流算法分析羔羊站立过程特征，结合两种算法提取的连续帧之间母羊舔羔和羔羊站立的时间和频次，采用Pearson进行相关性分析，采用单因素方差分析探究舔羔对羔羊站立的影响。【结果】 ①YOLOv5s算法对舔羔姿态识别的平均精度为97.90％，羔羊躺卧姿态识别的平均精度为90.70％，羔羊尝试站立姿态识别的平均精度为82.60％，羔羊站立姿态识别的平均精度为86.30％。②通过Farneback 光流算法分析羔羊站立过程特征，能有效展示羔羊从躺卧到站立过程中的运动轨迹和方向变化。③舔羔时间最长为1 462.00 s，最短为0 s，平均值为410.10 s；羔羊初次尝试站立时间在284.00～1 195.00 s之间，平均值为529.25 s；初次成功站立时间为695.00～2 921.00 s，平均值为1 464.35 s；尝试站立时间为323.00～1 785.00 s，平均值为930.10 s；羔羊尝试站立频次最小11.00次，最大46.00次，平均值23.60次。④Con组羔羊尝试站立次数明显高于Lick组，Con组和Lick组羔羊每次尝试站立持续时间随着尝试次数的增加呈现增加趋势。⑤舔羔总时长与羔羊尝试站立时长、羔羊初次成功站立时长、羔羊尝试站立频次呈显著负相关(P＜0.05)。⑥Lick组羔羊出生后至初次尝试站立时间与Con组羔羊无明显差异(P＞0.05)；Lick组羔羊尝试站立时间显著短于Con组羔羊(P＜0.05)；Lick组羔羊初次成功站立的时间显著短于Con组羔羊(P＜0.05)；Lick组羔羊尝试站立次数显著少于Con组羔羊(P＜0.05)。【结论】 YOLOv5s模型能够准确识别母羊舔羔和羔羊站立姿态，Farneback光流算法可以很好地分析羔羊站立过程特征，而母羊舔羔时间越长可以促使羔羊具有较高的活力，使羔羊尽快成功完成站立。

关键词: YOLOv5s模型; Farneback光流算法; 母羊舔羔; 羔羊站立

Abstract: 【Objective】 The intelligent detection of ewes licking lambs and the standing postures of lambs,the characteristics of the lamb standing process as well as the influence of ewes licking lambs on the standing of lambs were intended to be studied in this experiment. 【Method】 Videos of the licking behavior of 30 ewes and the standing process of lambs were collected to train the YOLOv5s model.20 multiparous single-fetus Saanen dairy goats and their newborn lambs were selected.Those with the licking time of ewes less than 120 s were divided into control group (Con group),and those with the licking time of ewes more than 120 s were divided into licking group (Lick group),with 10 in each group.The licking behavior of ewes and the standing postures of lambs were recognized based on the trained YOLOv5s algorithm.The characteristics of the standing process of lambs were analyzed by relying on the Farneback optical flow algorithm.The time and frequency of the ewes’ licking behavior and the lambs’ standing between consecutive frames were extracted by combining the two algorithms.Pearson correlation analysis was adopted,and One-Way ANOVA was used to explore the effect of lamb licking on lamb standing. 【Result】 ①The average precision of the YOLOv5s algorithm for identifying the licking posture was 97.90%.The average precision for identifying the lying posture of lambs was 90.70%.The average precision for identifying the posture of lambs attempting to stand was 82.60%.And the average precision for identifying the standing posture of lambs was 86.30%.②Through the analysis of the characteristics of the lamb standing process by the Farneback optical flow algorithm,the motion trajectory and direction changes of lambs during the process from lying down to standing could be effectively demonstrated.③The longest licking time was 1 462.00 s,the shortest was 0 s,with an average of 410.10 s.The time when lambs first attempted to stand ranged from 284.00 to 1 195.00 s,with an average of 529.25 s.The time when lambs first successfully stood ranged from 695.00 to 2 921.00 s,with an average of 1 464.35 s.The time for attempting to stand ranged from 323.00 to 1 785.00 s,with an average of 930.10 s.The minimum number of times that lambs attempted to stand was 11.00,the maximum was 46.00,with an average of 23.60 times.④The number of attempts of lambs in Con group to stand was significantly higher than that of lambs in Lick group.The duration of each attempt to stand by lambs in both Con and the Lick groups showed an increasing tendency as the number of attempts increased.⑤The total duration of ewes licking lambs was significantly negatively correlated with the duration of lambs’ attempts to stand,the duration of lambs’ first successful standing and the frequency of lambs’ attempts to stand (P＜0.05).⑥There was no significant difference in the time from birth to the first attempt to stand between lambs in Lick group and those in Con group (P＞0.05).The time for lambs in Lick group to attempt to stand was significantly shorter than that of lambs in Con group (P＜0.05).The time for lambs in Lick group to first successfully stand was significantly shorter than that of lambs in Con group (P＜0.05).The number of attempts of lambs in Lick group to stand was significantly less than that of lambs in Con group (P＜0.05). 【Conclusion】 The YOLOv5s model could accurately identify the postures of ewes licking lambs and the standing postures of lambs,and the Farneback optical flow algorithm could well analyze the characteristics of the lamb standing process.Moreover,the longer the licking time of ewes was,the higher the vitality of lambs would be,which could help lambs successfully stand up as soon as possible.

Key words: YOLOv5s model; Farneback optical flow algorithm; ewe licking; lamb standing

中图分类号:

S827

王国山, 宋懿峰, 安晓萍, 王园, 刘娜, 王文文, 齐景伟. 母羊舔羔与新生羔羊站立姿态的识别及关联分析[J]. 中国畜牧兽医, 2025, 52(8): 3753-3766.

WANG Guoshan, SONG Yifeng, AN Xiaoping, WANG Yuan, LIU Na, WANG Wenwen, QI Jingwei. Identification and Correlation Analysis of Ewe Licking and Standing Posture in Newborn Lambs[J]. China Animal Husbandry & Veterinary Medicine, 2025, 52(8): 3753-3766.

参考文献

[1] 罗茜.浅谈奶山羊养殖过程中的生物安全管理[J].中国动物保健,2023,25(8):79-80. LUO Q.A brief discussion on biosecurity management in the dairy goat farming process[J].China Animal Health，2023，25(8)：79-80.(in Chinese)
[2] SAWALHA M R，CONINGTON J，BROTHERSTONE S, et al.Analyses of lamb survival of Scottish Blackface sheep[J].Animal，2007，1(1):151-157.
[3] BROWN D J，FOGARTY N M，IKER C L，et al.Genetic evaluation of maternal behaviour and temperament in Australian sheep[J].Animal Production Science，2016,56(4)：767-774.
[4] DODD L C，PITCHFORD S W，EDWARDS H E J,et al.Measures of behavioural reactivity and their relationships with production traits in sheep:A review[J].Applied Animal Behaviour Science，2012，140(1-2)：1-15.
[5] 王兰萍，耿荣庆，洪键，等.绵羊的母性行为及其研究进展[J].家畜生态学报,2012,33(3)：116-120. WANG L P，GENG R Q，HONG J,et al.Maternal behavior of sheep and its research progress[J].Journal of Domestic Animal Ecology，2012,33(3)：116-120.(in Chinese)
[6] RILEY D G,CHASE JR C C,OLSON T A,et al.Genetic and nongenetic influences on vigor at birth and preweaning mortality of purebred and high percentage Brahman calves[J].Journal of Animal Science，2004,82(6):1581-1588.
[7] DWYER C.The welfare of the neonatal lamb[J].Small Ruminant Research,2007,76 (1):31-41.
[8] 韩成全，张鹏宇，王慧，等.哺乳动物母性行为影响因素[J].生态学报，2024，44(8)：3107-3120. HA C Q，ZHANG P Y，WANG H,et al.Factors influencing maternal behavior in mammals[J].Acta Ecologica Sinica，2024,44(8)：3107-3120.(in Chinese)
[9] 王政,宋怀波,王云飞，等.奶牛运动行为智能监测研究进展与技术趋势[J].智慧农业(中英文),2022,4(2):36-52. WANG Z，SONG H B，WANG Y F,et al.Research progress and technology trend of intelligent morning of dairy cow motion behavior[J].Smart Agriculture，2022，4 (2):36-52.(in Chinese)
[10] 张译方.浅谈计算机网络技术在畜牧业中的应用[J].福建农业，2015,8:172. ZHANG Y F.A brief discussion on the application of computer network technology in animal husbandry[J].Fujian Agriculture,2015,8：172.(in Chinese)
[11] MA S F，ZHANG Q R，LI T F,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.
[12] MAN C，HONGBO Y，QIFAN W,et al.Application of deep learning in sheep behaviors recognition and influence analysis of training data characteristics on the recognition effect[J].Computers and Electronics in Agriculture，2022，198:107010.
[13] 宋怀波，李通，姜波，等.基于Horn-Schunck光流法的多目标反刍奶牛嘴部自动监测[J].农业工程学报，2018,34(10)：163-171. SONG H B，LI T，JIANG B，et al.Automatic detection of multi-target ruminate cow mouths based on Horn-Schunck optical flow algorithm[J].Transactions of the Chinese Society of Agricultural Engineering，2018，34(10)：163-171.(in Chinese)
[14] 宋怀波，吴頔华，阴旭强，等.基于Lucas-Kanade稀疏光流算法的奶牛呼吸行为检测[J].农业工程学报，2019，35(17)：215-224. SONG H B，WU D H，YIN X Q,et al.Respiratory behavior detection of cow based on Lucas-Kanade sparseoptical flow algorithm[J].Transactions of the Chinese Society of Agricultural Engineering，2019，35(17)：215-224.(in Chinese)
[15] 吴頔华.基于视频分析的奶牛呼吸行为检测方法研究[D].杨凌：西北农林科技大学，2021. WU D H.Detection of dairy cow’srespiratory behavior basedon video analysis[D].Yangling：Northwest A & F University，2021.(in Chinese)
[16] GÜL S，GÖRGÜLÜ Ö，KESKIN M,et al. Maternal behaviour of Awassi sheep and behaviour of thelambs during the first hour after parturition[J].Turkish Journal of Veterinary and Animal Sciences，2017，41(6):741-747.
[17] DWYER M C，CONINGTON J，CORBIERE F,et al.Invited review:Improving neonatal survival in small ruminants:Science into practice[J].Animal,2015,10(3)：449-459.
[18] RAMIREZ A，QUILES A，HEVIA M L,et al.Behaviour of the Murciano-Granadina goat during the first hour after parturition[J].Applied Animal Behaviour Science，1998，56(2-4)：223-230.
[19] MADANI T，ALOUCHE L，SAFFIDINE N, et al.Maternal and neonatal behaviors of Ouled Djellal sheep breed and their effects on production parameters[J]. Small Ruminant Research,2013,114(1)：46-50.
[20] DWYER C.Behavioural development in the neonatal lamb:Effect of maternal and birth-related factors[J].Theriogenology，2003,59(3-4)：1027-1050.
[21] LIDFORS L M.Behavioural effects of separating the dairy calf immediately or 4 days post-partum[J].Applied Animal Behaviour Science，1996，49(3)：269-283.