PCB缺陷检测数据集

印刷电路板（PCB）瑕疵数据集由北京大学发布，其中包含1386张图像以及6种缺陷（缺失孔，鼠标咬伤，开路，短路，杂散，伪铜），用于检测，分类和配准任务。我们选取了其中适用与检测任务的693张图像，下载内容包括JSON和VOC格式版。

The Printed Circuit Board (PCB) Defect Dataset was released by Peking University. It contains 1386 images covering six types of defects: missing holes, mouse bites, open circuits, short circuits, stray copper, and pseudo copper, and is designed for detection, classification and registration tasks. We selected 693 images suitable for the detection task from this dataset, and the downloaded content includes files in JSON and VOC formats.

搜集汇总

背景与挑战

背景概述

该数据集是一个用于PCB缺陷检测的数据集，采用JSON格式存储，大小为960.67MB，更新于2024年8月。它采用Apache License 2.0开源许可证，支持通过ModelScope SDK或GIT Clone方式下载，便于研究和应用集成。