<![CDATA[The investigation focuses on using deep learning techniques to design a testing camera system and identify welding-related issues in outer pipes. The system uses an RGB camera, an ESP32 microcontroller, a DC motor with a rotary encoder, an LED strip, and a small computer that has a graphical user interface. This prototype was made to help with automating the inspection of carbon steel pipe welding, focusing on identifying and grouping different types of defects like porosity, undercut, and too much reinforcement. The results show that the system works in real time, achieving an average of 10.35 frames per second, taking 0.21 seconds to detect something, and having a confidence level of 97.13%. The built-in GUI shows the original image, the edge detection results, and the cropped ROI area, along with motor control and an LED strip. Mechanically, the prototype demonstrates sufficient stability for laboratory research; however, for industrial-scale research purposes, adjustments to materials and dimensions are required. This further develops an automated inspection system based on digital image processing and is designed to be safe, efficient, and compliant with safety standards.]]>
Copyrights © 2026
