Development Of An Artificial Vision Algorithm for The Recognition of Preventive Traffic Signs

León León R.A.; Berrios Barreto E.R.; Haro Reyes J.C.; León Montero J.A.; Tandaypan Ruiz E.A. · CISCI - Conferencia Iberoamericana en Sistemas, Cibernetica e Informatica (2024) · DOI: 10.54808/cisci2024.01.483

Resumen

The research aims to recognize preventive traffic signs accurately through deep learning specifically convolutional neural network under the detection model YOLO v8 (You Only Look Once, version 8) using libraries such as, OpenCV, ultralytics, numpy, a laptop with 16 GB of RAM, NVIDIA Geforce GTX graphics card and Webcam. The code development in Roboflow segments 233 images into 3 sets: training, validation and testing. Then, the code is exported in YOLOv8 format. Subsequently, the code in Yolov8m, is trained, tested and validated in an image review for 100 epochs using Google Collab. Finally, the final accuracy of the model was 95%. © CISCI - Simposium Iberoamericano en Educacion, Cibernetica e Informatica 2024.

Instituciones: Universidad Privada del Norte