Solid waste classifier for the i.e. Juan XXIII from the municipality of Algeciras with the application of Machine Learning

Abstract

The objective of this research is to develop, through the implementation of Artificial Intelligence (AI), a solid waste classifier for the Juan XXIII Educational Institution of the Municipality of Algeciras Huila, in order to generate an attractive user experience that sensitizes users. students and the entire educational community about the importance of recycling and classifying solid waste, according to the new color codes provided for in resolution 2184 of 2019 (Ministry of the Environment, 2019). The methodology developed is of the applied type because it is focused on the solution of a problem with the application of knowledge and qualitative, because it integrates a bibliographic review on the context of environmental problems in Colombia and the use of immersive technologies such as AI; in turn, it was divided into the following phases: documentary stage, design, development, implementation and validation. As a result, a neural network or Machine Learning model (automatic learning) was obtained with a data set of three classes, each with an average of 800 samples taken from the prototype and downloaded from Google's free image banks. The model was made using the Google Teachable Machine lab, the tensorflow open source and the P5.js editor and obtained an accuracy of 89.63%. The application runs online using Google drive services as data storage and its innovation lies in the implementation of colored emoticons that visually generates an interaction with the user and shows him in which container the waste shown in front of the camera goes. Its implementation was carried out at the Juan XXIII Educational Institution under a demonstration event and was well accepted. In the future, it is expected that the project can be implemented in a fixed manner and contribute to the care of the environment.