Design of an Artificial Vision System for the Measurement of Cut Aluminium Profiles with YOLOv5 Convolutional Neural Networks

Lecaros D.; Paraguay A.; Figueroa J. · International Conference on Electrical, Computer, Communications and Mechatronics Engineering, ICECCME 2025 (2025) · DOI: 10.1109/iceccme64568.2025.11277773

Resumen

Aluminium profiles are essential components in the manufacture of facades, doors, partitions and windows, which require precise measurements to ensure their correct installation. This work presents the design of an automated artificial vision system to accurately measure the cuts of aluminum profiles using YOLOv5 convolutional neural networks. The system captures realtime images, processes them using grayscale conversion, Gaussian filtering, edge detection, and applies a trained object detection model. The proposed system reached 99% accuracy for profiles larger than 15 cm 97% for medium-sized profiles, and 92% for smaller ones, demonstrating suitability for typical industrial requirements.

Instituciones: Universidad Peruana de Ciencias Aplicadas