Development of an Algorithm using Convolutional Neural Networks for the classification of solid waste in universities

León León R.A.; Barreto Espejo K.D.; Cabanillas Vigo K.J.; Muñoz Chacón J.S.; Peña Martinez S.A.; Tafur Sepúlveda A.E. · CISCI - Conferencia Iberoamericana en Sistemas, Cibernetica e Informatica (2024) · DOI: 10.54808/cisci2024.01.505

Resumen

This research work put into practice the use of an algorithm using convolutional neural networks to carry out a correct classification of solid waste in universities. For this purpose, the Python programming language was used, which made it possible to easily handle the programming language of the YOLOv8 neural network, which is the most advanced model of the YOLO series. On the other hand, in its development the algorithm works with the information of a Dataset of 1203 images collected in Roboflow, which is used in the training stage, where the network is dedicated to learn and then manage to detect the type of residue, it should be noted that the model works at a confidence level of 0.94. Finally, the results indicate that the accuracy value of the model to classify the type of waste is 84.6% for cardboard, 97.1% for wood, 94.2% for paper and 96.4% for plastic; achieving an overall accuracy of 90.6%. Thus, it can be affirmed that the coding and training of the YOLOv8 neural network was successful, allowing the algorithm to effectively detect and classify the type of waste (plastic, paper, cardboard and wood) present in the universities. © CISCI - Simposium Iberoamericano en Educacion, Cibernetica e Informatica 2024.

Instituciones: Universidad Privada del Norte