Development of machine based on computer vision and artificial intelligence for selection of sugar cane seedlings

• Grant number: 17/07731-1
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE
• Duration: December 01, 2017 - November 30, 2019
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Convênio/Acordo: FINEP - PIPE/PAPPE Grant
• Principal Investigator: Fernando Paes Lopes
• Grantee: Fernando Paes Lopes
• Host Company: Mvisia Desenvolvimentos Inovadores Ltda
• CNAE: Fabricação de máquinas e equipamentos para a agricultura e pecuária, exceto para irrigação ; Desenvolvimento e licenciamento de programas de computador customizáveis
• City: São Paulo

Abstract

Motivated by the successful implementation of process PIPE FAPESP Phase I (number 2012/50974-9), and also by the good progress made with process PIPE FAPESP Phase II (number 2015/08706-5), the leading researcher and his team submit the current research project aiming to adapt the previously developed technology to the sugarcane market, which has a great importance for the national economy. Brazil is, currently, the biggest producer of sugarcane in the world, a sector that contributes with more than R$40 billions a year, accounting for about 2.35% of the country's GDP (1). The industry is one of the main job generators in the country, employing 3.6 million people directly and indirectly, and gathering more than 72,000 agricultors (2). The industry has been struggling for the past few years, mainly due to an imbalance between demand and supply. In the given context, raising the sugarcane producers' level of productivity is necessary for the business sustainability. The MPB (pre-planted seedlings, mudas pré-brotadas in Portuguese) system is an example of innovation that aims to raise the industry's productivity level. The system consists of a quick and efficient multiplication for the production of sugarcane seedlings. This new model optimizes the use of the planted area and of the raw material, besides reducing flaws, plagues and diseases in the process. The main issue of the MPB system is the high cost of labour force, since the activities associated with seedlings' production and quality control are completely manual. Automatizing the seedling quality analysis before the seedling is planted, one of the hardest and most expensive tasks done manually, is the goal of this project. Based on knowledge and experience from previous projects, a proof-of-concept of a machine based on computer vision and artificial intelligence capable of identifying the quality of the sugarcane seedling and selecting it as "able" or "not able" for cultivation has been developed. The goal of the current project is, therefore, the technical enhancement of the already developed solution, converting the proof-of-concept into a commercial product besides developing a plan to insert technology as a commercial product. (AU)