Printed Circuit Board Defect Detection using YOLOv7

Narasimhan, Madhusudan (2022) Printed Circuit Board Defect Detection using YOLOv7. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration manual) Download (807kB) | Preview

Abstract

This paper proposes using the YOLOv7 algorithm to identify defects on Printed Circuit Boards (PCBs) during the manufacturing process. Traditional methods for defect identification are costly and often result in false alarms. YOLOv7 is the fastest known object detection algorithm, making it well-suited for the high-demand environment of PCB manufacturing. The paper also discusses data augmentation techniques performed and performance improvements compared to current state-of-the-art object detection methods. The model building towards building a robust model was performed in five stages which included various levels of data augmentation and multi-stage fine tuning on the model. After considerable iterations the model built was performing well and the mAP @ 0.5 was recorded to be 95.8% with overall precision of the model at 98% percent and recall at 92%.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Nayak, Prashanth UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Algebra > Algorithms > Computer algorithms T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Electricity Supply
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Tamara Malone
Date Deposited:	23 May 2023 15:34
Last Modified:	23 May 2023 15:34
URI:	https://norma.ncirl.ie/id/eprint/6625

Actions (login required)

View Item