Traits Expansion and Storage of Soybean Phenotypic Data in Computer Vision-Based Test

Phenotypic traits of crops are an important basis for cultivating new crop varieties. Breeding experts expect to use artificial intelligence (AI) technology and obtain many accurate phenotypic data at a lower cost for the design of breeding programs. Computer vision (CV) has a higher resolution than human vision and has the potential to achieve large-scale, low-cost, and accurate analysis and identification of crop phenotypes. The existing criteria for investigating phenotypic traits are oriented to artificial species examination, among these are a few traits type that cannot meet the needs of machine learning even if the data are complete. Therefore, the research starts from the need to collect phenotypic data based on CV technology to expand, respectively, the four types of traits in the “Guide to Plant Variety Specificity, Consistency and Stability Testing: Soybean”: main agronomic traits in field investigation, main agronomic traits in the indoor survey, resistance traits, and soybean seed phenotypic traits. This paper expounds on the role of the newly added phenotypic traits and shows the necessity of adding them with some instances. The expanded traits are important additions and improvements to the existing criteria. Databases containing expanded traits are important sources of data for Soybean AI Breeding Platforms. They are necessary to provide convenience for deep learning and support the experts to design accurate breeding programs.