针灸机器人腧穴定位数据集

所有腧穴的初始位置均由至少两名具备资质的针灸医师，依据国家标准的腧穴定位规范（如 GB/T12346-2021）进行实地揣穴与交叉核对后，使用物理标签进行标记并通过搭载深度相机的机械臂移动来采集数据。对采集的原始图像，使用指定的标注工具如（如labelimg）将物理标签的位置精确标注为边界框，并以指定的格式（如 YOLO 格式）生成标签文件。The initial positions of all acupoints were determined by at least two qualified acupuncturists, who performed on-site acupoint palpation and cross-verification in compliance with national standard specifications for acupoint nomenclature and location (e.g., GB/T 12346-2021). Subsequently, the positions were marked with physical markers, and data collection was conducted via a robotic arm equipped with a depth camera during its movement. For the collected raw images, designated annotation tools such as LabelImg were used to accurately annotate the positions of the physical markers as bounding boxes, and annotation files were generated in specified formats such as the YOLO format.

搜集汇总

背景与挑战

背景概述

该数据集是由天津中医药大学提供的大规模腧穴定位数据集，包含60643条数据，覆盖多个腧穴的精确坐标信息，适用于针灸机器人视觉识别算法的训练与测试。数据由专业医师依据国家标准标注，采集环境贴近临床场景，旨在解决腧穴标注数据稀缺和提升模型在复杂条件下的鲁棒性，支持针灸智能装备研究和教学应用。