This is the authors' abstract. We don't add key points for this paper.
From the past few years, the Printed Circuit Boards (PCBs) played a significant role in electronic devices and served as crucial components among different industries like medical, automotive, industrial and other sectors. Traditional approaches for fault detection in PCB had faced challenges like reliance on manual inspection and inefficiency in handling complex PCB layouts. Therefore, this research proposes Deep Learning-based Atrous Spatial Pyramid Pooling version 3+ (DeepLabv3+) for fault Detection in PCB Manufacturing. Initially, data is collected from PCB dataset which consists of s 2500 images in three conditions. Then, Oriented fast and Rotated Brief (ORB) algorithm is utilized for extracting features from input images. Next, Random Sample Consensus (RANSAC) algorithm calibrates input images for matching size and orientation of template data. Finally, faults are detected by using proposed DeepLabv3+ which uses architecture of encoder-decoder with Atrous Spatial Pyramid Pooling (ASPP) to accurately segment, identify defects on the PCB surface. The proposed DeepLabv3+ achieved better results in conditions of three different lighting with an accuracy (98%), and 12 Frames Per Second (FPS) of detection speed respectively when compared with existing Convolutional Neural Network (CNN).
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Yarram et al. (2025) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: