A1265
Title: Human-centered integration of AI-based phenotyping in modern breeding pipelines
Authors: Wei Guo - The University of Tokyo (Japan) [presenting]
Abstract: Integrating artificial intelligence (AI) and image-based phenotyping is revolutionizing how plant breeding programs collect, analyze, and utilize trait data under real-world conditions. These technologies enable non-destructive, high-throughput assessment of crop growth, morphology, and stress responses, offering powerful support for genotype selection and environmental adaptation. However, meaningful adoption in breeding contexts requires a human-centered approach that ensures interpretability, usability, and actionable outcomes. Recent efforts in developing AI-powered phenotyping pipelines designed for direct application in breeding programs are presented. We demonstrate how trait data extracted via deep learning and UAV-based imaging are integrated with genotype and environmental variables through robust modeling workflows. Emphasis is placed on spatial correction, trait calibration, and the alignment of analytical outputs with breeder decision-making needs. By combining technical innovation with human usability, our work contributes to building scalable, flexible phenotyping systems that accelerate genetic improvement, support climate-resilient cultivar development, and foster stakeholder confidence in AI-driven tools.
