High throughput phenotyping and machine learning for selection of soybean genotypes resistant to stink bug complex in different maturity groups

• Grant number: 25/11161-2
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: July 01, 2025
• End date: June 30, 2026
• Field of knowledge: Agronomical Sciences - Agronomy
• Principal Investigator: José Baldin Pinheiro
• Grantee: José Tiago Barroso Chagas
• Host Institution: Escola Superior de Agricultura Luiz de Queiroz (ESALQ). Universidade de São Paulo (USP). Piracicaba , SP, Brazil
• Associated research grant: 24/15430-5 - Machine learning-based multi-modal data fusion and growth modeling for soybean production improvement, AP.PFPMCG.TEM

Abstract

Soybeans are a species of strategic importance in Brazil, being the most economically important crop in the country. The stink bug complex stands out among the main pests that attack crops, which causes important reductions in grain production and seed quality. The most efficient strategy to overcome the problem is to obtain resistant cultivars, and for this, high-precision phenotyping has shown promising results. Therefore, the present study aims to develop a high-precision phenotyping platform for identifying superior lines, taking into account the response to infestation by the soybean stink bug complex and the crop cycle. Thus, two populations will be evaluated, the first with 300 and the second with 250 individuals, conducted under two management strategies. The genotypes in different crop phenological stages will be phenotyped in the field through aerial images and on the ground with RGB sensors. Grain yield and seed quality will be measured in the laboratory. Using two deep machine learning approaches, the data will be used as explanatory variables for developing computer vision classification models and productivity prediction. The developed platform can be used in genetic improvement programs aimed at early selection for strains resistant to the soybean stink bug complex. In addition, applying prediction models in soybean crops can help define advantageous real-time pest control strategies, leading to the rational application of inputs and reduction of environmental and crop production costs. (AU)