Abstract: Quick and accurate recognition of individual cattle is of great significance for disease prevention and control, genetic improvement of varieties, traceability of dairy products and meat products, and improvement of agricultural false insurance claims. Traditional cattle identification methods such as   hot iron branding, ear tattooing, ear tagging and radio frequency identification, are subject to equipment loss, duplication, fraud, animal welfare security, monitoring cost and distance challenges. Instead, based on biometrics, non-contact identification is a new trend in cattle identification due to its uniqueness, invariance, low cost, easy operation and high animal welfare. This paper reviewed several research progress based on non-contact cattle identification, mainly focused on the latest achievements of cattle face recognition, and discussed current challenges faced in the practical application of cattle face recognition. The design and future prospects of the application of deep learning in cattle recognition were proposed in this paper.

Key words: individual identification, non-contact, biometrics, face recognition, deep learning

摘要： 快速精准确定牛个体身份对疾病防控、品种遗传改良、奶制品和肉制品质量溯源以及改善农业假保险索赔等方面具有重要意义。传统的牛个体识别使用诸如烙印、耳纹、耳标和无线射频识别等方法，易遭受设备损失/工作重复、标记欺诈、动物福利安全以及监测成本和距离等方面的挑战；而基于生物特征的非接触识别由于其独特性、不变性、低成本易操作以及动物福利高，成为牛身份识别的新趋势。主要介绍了几种基于非接触式的牛身份识别的研究进展，重点关注牛脸识别的最新成果，讨论当前牛脸识别在实际应用中面临的挑战，在此基础上对深度学习在牛脸身份识别研究中的应用进行了设计构思与展望。

关键词: 个体身份, 非接触, 生物特征, 牛脸识别, 深度学习