Automated Vision Inspection Based IC Component Locator Using Deep Learning

Mohamed, Ibrahim Soliman (2019) Automated Vision Inspection Based IC Component Locator Using Deep Learning. Project Report. UTeM, Melaka. (Submitted)

Text (24 Pages) Automated vision inspection based IC component locator using deep learning.pdf - Submitted Version Download (289kB)

Abstract

With the coming of the era of industrial revolution 4.0, manufacturers produce high-tech products. As the production process is refined, inspection technologies become more important. Specifically, the inspection of a printed circuit board (PCB), which is an indispensable part of electronic products, is an essential step to improve the quality of the process and yield. Image processing techniques are utilized for inspection, but there are limitations because the backgrounds of images are different, and the kinds of component shape and size parameters are normally various. In order to overcome these limitations, methods based on machine learning and deep learning have been developed recently. In this project, I have developed an IC components locator software to help in inspection process, this software is relying on 2 model of most popular object detection on deep learning field (Yolo V3 and Faster RCNN), I have trained bot models and preformed a comparison between their results in term of mAP, loss, inference time and training time. Yolo V3 and Faster RCNN have been trained on a filtered open source dataset of PCB that contains 163 Images, an annotation and augmentation tool has been developed in purpose of increasing the amount of our dataset. Finally, OPENVINO toolkit has used for optimization process and infer both models on various Intel CPU to run our deep learning network on edge.

Actions (login required)

View Item