A YOLO Annotated 15-class Ground Truth Dataset for Substation Equipment

This dataset contains 7539 images of electric substations with 213566 annotated objects. The images were obtained using different cameras, ranging from cellphone cameras, panoramic aerial cameras, and stereo FLIR cameras, including cameras mounted on an Autonomous Guided Vehicles (AGV). A total of 15 classes of objects were identified in this dataset. These are the classes and their number of instances: Open blade disconnect switch: 1117Closed blade disconnect switch: 26068Open tandem disconnect switch: 4263Closed tandem disconnect switch: 5402Breaker: 4803Fuse disconnect switch: 1925Glass disc insulator: 13803Porcelain pin insulator: 114578Muffle: 8128Lightning arrester: 8788Recloser: 8059Power transformer: 2391Current transformer: 9293Potential transformer: 2904Tripolar disconnect switch: 2044All images in this dataset were collected from a single electrical distribution substation in Brazil over a period of two years. The images were captured at various times of the day and under different weather and seasonal conditions, ensuring a diverse range of lighting conditions for the depicted objects. A team of experts in Electrical Engineering curated all the images to ensure that the angles and distances depicted in the images are suitable for automating inspections in an electrical substation.The file structure of this dataset contains the following directories and files: misc: This directory contains 4030 images collected manually during the morning period, ranging from 8h00 up to 12h00;misc/labels: This subdirectory contains the 4030 YOLO annotated .txt files for the misc directory;agv_day: This directory contains 2270 images collected using an automated guided vehicle during daytime, ranging between 8h00 and 10h00 during morning and in the afternoon between 13h00 and 17h00;agv_day/labels: This subdirectory contains the 2270 YOLO annotated .txt files for the agv_day directory;agv_night_light: This directory contains 899 images collected using an automated guided vehicle during nighttime, where artificial lighting is present, ranging from 20h00 up to 21h00;agv_night_light/labels: This subdirectory contains the 899 YOLO annotated .txt files for the agv_night_light directory;agv_night_dark: This directory contains 340 images collected using an automated guided vehicle during nighttime, without artificial light sources, ranging from 20h00 up to 21h00;agv_night_dark/labels: This subdirectory contains the 340 YOLO annotated .txt files for the agv_night_dark directory;classes.txt: This text file lists the 15 classes indexed in the order used in the annotations.The dataset aims to support the development of computer vision techniques and deep learning algorithms for automating the inspection process of electrical substations. We expect it to be useful for researchers, practitioners, and engineers interested in developing and testing object detection models for automating inspection and maintenance activities in electrical substations.

本数据集共收录7539张变电站场景图像，总计标注213566个目标实例。图像采集自多类相机设备，包括手机相机、全景航空相机、立体FLIR相机，以及搭载于自动导引车（AGV）上的相机。本数据集共标注15类目标对象，各类别及其实例数量如下：分闸刀闸式隔离开关（Open blade disconnect switch）：1117；合闸刀闸式隔离开关（Closed blade disconnect switch）：26068；分闸双柱式隔离开关（Open tandem disconnect switch）：4263；合闸双柱式隔离开关（Closed tandem disconnect switch）：5402；断路器（Breaker）：4803；熔断器式隔离开关（Fuse disconnect switch）：1925；玻璃盘形绝缘子（Glass disc insulator）：13803；瓷针式绝缘子（Porcelain pin insulator）：114578；消声罩（Muffle）：8128；避雷器（Lightning arrester）：8788；重合器（Recloser）：8059；电力变压器（Power transformer）：2391；电流互感器（Current transformer）：9293；电压互感器（Potential transformer）：2904；三极隔离开关（Tripolar disconnect switch）：2044。

本数据集所有图像均采集自巴西境内某一座配电变电站，采集周期长达两年。图像拍摄于一日内的不同时段，涵盖多样的天气与季节条件，确保了场景中目标对象的光照环境具备丰富多样性。电气工程领域专家团队对全部图像进行了筛选与标注校验，确保图像的拍摄角度与拍摄距离适配变电站巡检自动化的应用场景。

本数据集的文件结构包含如下目录与文件：
misc目录：收录4030张手动采集的晨间图像，拍摄时段为8:00至12:00；misc/labels子目录：存放对应misc目录下4030张图像的YOLO（You Only Look Once）格式标注文本文件；
agv_day目录：收录2270张由自动导引车采集的日间图像，拍摄时段涵盖上午8:00至10:00，以及下午13:00至17:00；agv_day/labels子目录：存放对应agv_day目录下2270张图像的YOLO格式标注文本文件；
agv_night_light目录：收录899张由自动导引车采集的夜间人工照明场景图像，拍摄时段为20:00至21:00；agv_night_light/labels子目录：存放对应agv_night_light目录下899张图像的YOLO格式标注文本文件；
agv_night_dark目录：收录340张由自动导引车采集的夜间无人工照明场景图像，拍摄时段为20:00至21:00；agv_night_dark/labels子目录：存放对应agv_night_dark目录下340张图像的YOLO格式标注文本文件；
classes.txt文本文件：按照标注索引顺序，列出了本数据集的15类目标对象。

本数据集旨在支撑面向变电站巡检自动化的计算机视觉技术与深度学习算法研发工作。我们期望该数据集能够为致力于开发、测试变电站巡检与运维自动化目标检测模型的研究人员、行业从业者及工程师提供有效支撑。

搜集汇总

背景与挑战

背景概述

该数据集包含7539张变电站设备的图像，标注了15类共213566个对象，图像采集方式多样且覆盖不同环境条件，旨在支持变电站自动化检查的计算机视觉研究。