Published August 12, 2025 | Version 1.0.0

EPID: The Enfield PCB Inspection Dataset for Visual Defect Detection

Description

This dataset contains high-resolution images of printed circuit boards (PCBs) captured under controlled laboratory conditions for research on defect detection. It includes both undamaged boards and boards with artificially introduced defects (burned integrated circuits (ICs), melted IC pins, popped capacitors), documented step-by-step using a reproducible damage-planning procedure.

Each image is annotated with bounding boxes for integrated circuits (ICs) and capacitors in COCO format, accompanied by metadata detailing the damage type and sequence. Designed for training and benchmarking models in low-data-volume scenarios, the dataset supports applications in quality control, manufacturing, and computer vision research.

License: CC BY-SA 4.0
Related resources: Dataset description [link]

Files

EPID.zip

Files (992.3 MB)

Name Size
md5:e43ac4a858ef7190f0b0920dcd27e0e4
992.3 MB Preview Download

Additional details

Related works

Has part
Software: https://github.com/Fain-Tech/lowdata-pcb-nas (URL)
Is described by
Data paper: 10.5281/zenodo.16874934 (DOI)

Funding

UK Research and Innovation
ENFIELD: European Lighthouse to Manifest Trustworthy and Green AI 10094603

Dates

Submitted
2025-02-12

Software

Repository URL
https://github.com/Fain-Tech/lowdata-pcb-nas
Programming language
Python
Development Status
Active