SVCf – Sistema para Detecção de Pontos de Corte em Pedúnculos de Frutas: Estudo de Caso com a Mangifera Indica 1

Carlos Hairon Ribeiro Gonçalves; André Luis Albuquerque Pinheiro; Pedro Henrile Salvador; José Marquês Soares

doi:10.20906/CBA2022/3769

Carlos Hairon Ribeiro Gonçalves Instituto Federal de Educação Ciência e Tecnologia do Ceará, campus Fortaleza
André Luis Albuquerque Pinheiro Instituto Federal de Educação Ciência e Tecnologia do Ceará, campus Crato
Pedro Henrile Salvador Instituto Federal de Educação Ciência e Tecnologia do Ceará, campus Crato
José Marquês Soares Universidade Federal do Cerará, Depto. de Eng. de Teleinformática, Campus do PICI

DOI: https://doi.org/10.20906/CBA2022/3769

Keywords: Computer vision, Cut Point Location, HSV Space, Mango Identification, R-CNN

Abstract

Fruit harvest can be automated by low-cost robots integrating computer vision systems. For this purpose, the SVCf is a computer vision system appropriated to UAVs and collector robots based on Agriculture 4.0 technologies. Initially, mango fruit (Mangifera Indica) was chosen to validate the proposed model used at SVCf. This proposed model can also be generalized to other types of fruits. R-CNNs were used to identify the fruits and color segmentation in the HSV space to identify the peduncles. In terms of results, the accuracy is about 97% in the correct identification of cut points in the peduncle of fruits located in real trees.