Patch-Wise Automatic Segmentation for Real-Time PCB Inspection
Keywords: Automated Optical Inspection, Printed Circuit Board, Patch-Wise, Automatic Segmentation, YOLOv7
Abstract: Automated Optical Inspection (AOI) systems play a pivotal role in ensuring quality control during Printed Circuit Board (PCB) manufacturing. However, the current AOI systems necessitate manual setting of the region of interest (ROI) for all components. To address this, we propose a patch-based preprocessing technique, dividing high-resolution PCB images into small 1024 × 1024 pixel patches and employing the YOLOv7 segmentation model for real-time component ROI segmentation. Our method consistently delivered high accuracy across various PCB components, irrespective of background color, and demonstrated robust performance even with complex structures containing small components. It achieved impressive outcomes, with an average IoU, F1 score, pixel accuracy, and mAP of 0.8889, 0.9401, 0.9961, and 0.8255, respectively. Specifically, utilizing Feature Pyramid Network (FPN) and Path Aggregation Network (PAN) in YOLOv7's multi-resolution processing allowed us to accurately segment PCB components of various sizes and process them in real-time. This study underscores the potential of automating real-time component ROI segmentation in the PCB manufacturing process to enhance production speed and quality control.
Primary Area: applications to computer vision, audio, language, and other modalities
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 9260
Loading