Cattle side view and back view dataset

The dataset was collected from small villages in eastern Chinese Inner Mongolia Autonomous region, which is known as the hometown of Horqin cattle. The dataset contains side and back views of 72 cattle. Each cattle is accompanied by detailed annotations, including oblique body length, withers height, heart girth, hip length, as well as body weight among other crucial data points. At present, the number of image datasets that can be used to analyze cattle is very limited. In the public datasets, most of them contain multiple cows in each image, which is mainly for recognition tasks, and it is impossible to determine the detailed feature points of each cow to measure the body size. This data set allows us to perform more complex estimation tasks such as automated measurements and weight prediction. In the field of computer vision, utilizing this dataset can facilitate the construction of deep learning models to develop an automated livestock monitoring system. Improve the management efficiency and economic benefits of animal husbandry.

本数据集采集自中国内蒙古自治区东部的乡村区域，该区域是科尔沁牛（Horqin cattle）的原产地。本数据集包含72头牛的侧面与背面影像，每头牛均配有详细标注信息，涵盖体斜长、鬐甲高、胸围、尻长以及体重等多项关键数据指标。当前可用于牛只分析的影像数据集数量十分有限。现有公开数据集多为单张图像包含多头牛只的场景，此类数据集主要服务于识别任务，无法精准定位单头牛的细节特征点以开展体尺测量。本数据集可支持自动化体尺测量、体重预测等更为复杂的估算任务。在计算机视觉领域，依托该数据集能够助力深度学习模型的构建，进而开发自动化畜牧监测系统，提升畜牧业的管理效率与经济效益。