Related Experiment Video
Updated: Jun 20, 2025
11:34
Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
Published on: May 15, 2017
11.1K
A dataset for deep learning based detection of printed circuit board surface defect
Shengping Lv1, Bin Ouyang2, Zhihua Deng3
1School of Engineering, South China Agricultural University, No. 483, Wushan Road, Guangzhou, 510642, China. lvshengping@scau.edu.cn.
Scientific Data
|July 22, 2024
Summary
This study introduces DsPCBSD+, a large dataset for detecting printed circuit board (PCB) surface defects. This resource aims to accelerate deep learning model development for improved PCB quality control.
Area of Science:
- Materials Science
- Computer Science
- Electrical Engineering
Background:
- Printed circuit board (PCB) manufacturing involves potential surface defects impacting performance and requiring robust quality control.
- Current defect detection methods need enhancement for efficiency and accuracy in complex manufacturing environments.
Purpose of the Study:
- To develop a comprehensive and diverse dataset for training deep learning models for PCB surface defect detection.
- To categorize PCB surface defects systematically to improve model generalization.
Main Methods:
- Categorization of PCB surface defects into 9 distinct types based on morphology, cause, and location.
- Development of the DsPCBSD+ dataset, comprising 10,259 images with 20,276 manually annotated defects using bounding boxes.
Main Results:
- Creation of the DsPCBSD+ dataset, a significant resource for PCB surface defect research.
- The dataset covers 9 distinct defect categories, facilitating diverse model training.
Conclusions:
- The DsPCBSD+ dataset is openly accessible to promote research and advancements in deep learning-based PCB surface defect detection.
- Availability of this dataset is crucial for accelerating the development of more accurate and efficient automated quality control systems for PCBs.
