Dataset de clasificación de Residuos Sólidos Urbanos para Redes Neuronales Convolucionales.

Mirna Castro Bello; Víctor Manuel Romero Juárez; Jorge Fuentes Pacheco; Sergio Ricardo Zagal Barrera; Areli Bárcenas Nava; Diego Esteban Gutiérrez Valencia

doi:10.46377/dilemas.v12i3.4675

Mirna Castro Bello Tecnológico Nacional de México
Víctor Manuel Romero Juárez Tecnológico Nacional de México
Jorge Fuentes Pacheco Tecnológico Nacional de México
Sergio Ricardo Zagal Barrera Tecnológico Nacional de México
Areli Bárcenas Nava Tecnológico Nacional de México
Diego Esteban Gutiérrez Valencia Tecnológico Nacional de México

DOI: https://doi.org/10.46377/dilemas.v12i3.4675

Keywords: dataset, urban solid waste, convolutional neural networks, deep learning, classification

Abstract

Convolutional Neural Network (CNN) models require processed data that learns image patterns to avoid memorization. This research presents the creation of a dataset of 3,208 images for the classification of Urban Solid Waste (MSW) into organic and inorganic forms for CNN models. Processing was carried out using a three-phase methodology: 1. Dataset identification and selection: Kaggle and Github; 2. Dataset creation: image uniformity and color adjustments; and 3. Creation of the organic and inorganic waste dataset. The results obtained were the organic MSW datasets consisting of 1,574 images and 1,634 inorganic images. This will enable the training of Deep Learning models for binary MSW classification.