| Title |
Automated visual inspection system based on stereo vision / |
| Translation of Title |
Automatinės vizualinės apžiūros sistemos, pagrįstos stereo vizija, projektavimas. |
| Authors |
Elatroush, Ahmed Ibrahim Mosad Hassan |
| Full Text |
|
| Pages |
78 |
| Keywords [eng] |
stereo vision ; deep learning ; object detection ; automated visual inspection ; quality control |
| Abstract [eng] |
An automated visual inspection system incorporating computer vision was investigated, and a prototype for a proof of concept was developed in this final degree project. The extensive usage of deep learning in the 4th industrial revolution was the inspiring factor for implementing it as part of the system. A traditional computer vision method known as stereovision integrated with deep learning-based artificial intelligence to simplify the algorithm used for depth measurement. This technique can be used widely in various inspection systems in different industries. The industry of focus is the electronics industry, where precise small components have their depths of fit being measured. The proposed system is cost-effective. The deep learning-based neural network was deployed on the Nvidia Jetson Nano board, which operates on the Linux OS. Python programming language was used to develop the code, and an x-y translation stage was integrated into this prototype. The system incorporated the newly released Raspberry Pi HQ cameras that support interchangeable lenses, allowing for the mounting microscopic lenses. |
| Dissertation Institution |
Kauno technologijos universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2021 |
