Development of neural models and of random forests to classify soybeans, beans, chickpeas and corn from digital images

• Grant number: 22/11858-5
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: January 01, 2023
• End date: December 31, 2023
• Field of knowledge: Interdisciplinary Subjects
• Principal Investigator: Erica Regina Filletti Nascimento
• Grantee: Vinícius Liberato Erler
• Host Institution: Instituto de Química (IQ). Universidade Estadual Paulista (UNESP). Campus de Araraquara. Araraquara , SP, Brazil

Abstract

In this project, we propose the development of Artificial Neural Networks (ANNs) and Random Forests as alternative tools to classify legume grains from digital images. In particular, the color characteristics of the images of samples of beans, soybeans, corn and chickpeas will be used, as well as the apparent density of the grains, to create a database that will be used for training the models. The ability to learn through examples and to generalize the information learned is, without a doubt, the main attraction of problem solving through ANNs. Some of the advantages of Random Forests is that they can be used for both regression and classification and it is easy to visualize the relative importance they assign to each characteristic in their inputs, in addition to producing a good result with default values for their parameters.Due to these advantages, these two tools will be very useful for the study of the properties of the grains, being able to help in the control of the post-harvest quality for these products of great agro-industrial relevance.