Plant Phenotyping

Plant phenotyping studies how to measure plant structure, development, health, and yield-related traits, helping breeders and researchers compare varieties and understand responses to growing conditions. Common approaches use cameras and spectral sensors with CNNs or vision transformers for detection, segmentation, counting, and trait prediction, alongside 3D reconstruction from images or point clouds and time-series models for tracking growth. Work increasingly explores self-supervised and foundation models, synthetic data, and multimodal sensing to reduce annotation needs and transfer across crops, environments, and tasks; evaluations include field and greenhouse imagery, 3D plant datasets, and seasonal UAV surveys. A central challenge is making measurements accurate and biologically meaningful under occlusion, changing light, and domain shifts, while keeping systems efficient and validating their performance beyond the data on which they were trained.