奶牛姿态估计数据集

奶牛姿态估计数据集专为奶牛姿态估计与物体检测任务设计，基于 12 个标记关键点（包括鼻子、右眼、左眼、颈部（侧面）、左前蹄、右前蹄、左后蹄、右后蹄、主干网（侧面）、尾根、后姿势中心（靠近尾部中点）、腹部（侧卧式中心））进行标注；采用 COCO 形式，支持 YOLOv8 等模型，包含背景、图像、标签及 metadata.txt 文件，总文件数 472 个，图像格式有.jpg、.webp、.jpeg 等

The Cattle Pose Estimation Dataset is specifically developed for cattle pose estimation and object detection tasks. It is annotated with 12 labeled keypoints including nose, right eye, left eye, neck (lateral view), left front hoof, right front hoof, left rear hoof, right rear hoof, trunk (lateral view), root of tail, posterior posture center (near the midpoint of the tail), and abdomen (center in side-lying posture). The dataset follows the COCO format, supports models such as YOLOv8, and contains background files, images, labels and metadata.txt. There are 472 files in total, and the supported image formats include .jpg, .webp, .jpeg, etc.

搜集汇总

背景与挑战

背景概述

奶牛姿态估计数据集专为奶牛姿态估计与物体检测任务设计，基于12个标记关键点（如鼻子、眼睛、蹄部等）进行COCO格式标注，支持YOLOv8等模型训练。该数据集包含472个文件，包括多种图像格式，适用于农业领域的奶牛行为分析和健康监测等计算机视觉应用。