Automação no agronegócio de pequeno porte: protótipo para seleção de morangos com uso de visão computacional e inteligência artificial

Abstract

The cultivation of food through family farming has great representativeness in Brazilian. However, the availability of technological resources that are applicable and accessible to this demand is low, making the production be manually. Due to this fact, we studied the development of technology aimed at the small farmer, in particular the development of a prototype selector for the selection of strawberries. Computational Vision and artificial intelligence was used, trained to select strawberries based on their maturity, in order to attend the producers of São José dos Pinhais, metropolitan region of Curitiba, Paraná. For the implementation of the computer vision, the OpenCV library was used and for the artificial intelligence the TensorFlow library, both open source, was used. An accuracy of 98.72% in the selection of green and ripe strawberries was reached at the end of the prototype development, as well as a capacity to select 207.36 kg of strawberries every day.