Cow delivery detection method

The invention discloses a cow delivery detection method. The method comprises the steps that behavior videos of a cow farm and cows are collected through a camera; collecting temperature and exercise amount data of the cow through an ear tag sensor; analyzing the collected behavior video through a deep learning model, extracting a behavior sequence, and performing feature extraction on behavior sequence information, temperature and exercise amount data within specific time; a machine learning model is adopted for modeling, whether cows are parturient or not is analyzed through the model, a farmer is informed of a parturient signal in advance, corresponding preparation measures are taken, and the survival rate of calves is increased. 本发明公开了一种母牛分娩的检测方法，所述方法包括：通过摄像头采集牛场和母牛的行为视频；通过耳标传感器采集母牛的温度和运动量数据；通过深度学习模型对采集的行为视频进行分析，提取行为序列，对特定时间内的行为序列信息、温度和运动量数据进行特征提取；采用机器学习模型进行建模，通过模型分析母牛是否分娩，并将即将分娩的信号提前告知养殖户，做好相应的准备措施，提升牛犊的生存率。