Body-Weight Estimation of Plateau Yak with Simple Dimensional Measurement

Animal husbandry is the economic backbone in large areas around the world, especially in rural and under-developing countries. In the Sanjiangyuan region of China, the cultivation and depasturing of yak is one of the most representative industries and the main source of revenue for the shepherds. In yak-breeding, body weight is an essential index for evaluating the body (health) status of the livestock. However, as a pasture-raised large-body-size animal, it is not easy for the herdsmen to weigh them. In this paper, we attempt to investigate the feasibility to acquire yaks’ body-wight with easy effort, i.e., via just measuring some dimensional features with ordinary rulers. To achieve this, we collect large-scale data about body dimensional features and the body weight of the yaks, and use machine learning methods to get the model to estimate the body weight. Through in-depth evaluation and analysis, we find that the estimation method by utilizing Random Forest achieves the best performance, i.e., with the test precision of 0.95 (The test precision of LR and SVMs is 0.92, 0.7 respectively). Finally, we develop the PC software and Android App based on our method and freely issue to the herdsmen to use. Our research greatly facilitates herdsmen’s livestock production.