Multimodal AI Phenotyping for Sustainable Crop Production: Structure-to-Function Trait Quantification from Non-Destructive 2D and 3D Sensing

Sustainable crop production depends on accurately and continuously characterizing how crops grow, function, and respond to changing environments. Non-destructive sensors — RGB, depth, hyperspectral, fluorescence, thermal, LiDAR, and 3D reconstruction — now capture complementary morphological, physiological, and biochemical signals across organ, plant, and canopy scales. Yet most studies still process a single modality, target a single trait family, or stop at structural reconstruction without linking it to function. As a result, sensor outputs rarely translate into coherent, biologically interpretable trait sets that can inform sustainable crop management. Artificial intelligence offers the connective tissue: fusing heterogeneous streams, integrating structure with function, and producing reproducible trait estimates. Strengthening this multimodal, structure-to-function integration is now a key step in turning sensing technology into dependable phenotypic information for sustainable crop systems.This Research Topic aims to advance AI-driven multimodal phenotyping that jointly recovers morphological, physiological, and biochemical traits from non-destructive 2D and 3D sensing, with a particular focus on structure-to-function integration for sustainable crop production. We welcome contributions that develop sensor-fusion strategies, 2D and 3D reconstruction and segmentation methods, interpretable trait-extraction models, and reproducible workflows that link raw imaging, spectral, and point-cloud data to biologically meaningful crop traits in greenhouse, field, UAV, proximal, and robotic settings. Rather than focusing on stress event detection, autonomous decision-making, machinery control, or yield estimation — all well covered by neighboring collections in this section — this Topic emphasizes continuous trait quantification and biological interpretation of the morphology–physiology–biochemistry triad. The collection seeks to consolidate methodological progress that supports phenotype-informed crop management, varietal characterization, and resource-use efficiency in sustainable production systems.We welcome Original Research, Reviews, Methods, Technology and Code, Perspective, and Opinion articles on:· AI-based fusion of crop images, point clouds, spectra, fluorescence, and thermal signals for joint multimodal trait inference· 2D and 3D reconstruction, segmentation, and tracking pipelines that yield quantitative morphological traits at organ, plant, and canopy scales· Estimation of physiological and biochemical traits — water status, photosynthetic activity, pigments, nutrients, metabolites — from non-destructive multimodal sensing· Structure-to-function trait integration linking reconstructed plant geometry with spectral and fluorescence-derived functional readouts· Interpretable machine learning, deep learning, foundation models, and digital twins for mapping multimodal signals to biologically meaningful crop traits· Reproducible workflows, benchmark datasets, and cross-platform transferability of multimodal phenotyping pipelines across greenhouse, field, UAV, proximal, and robotic acquisition settings, in support of sustainable crop production-------We welcome contributions from researchers across the field, and we are glad to support you as you prepare your submission. If you would like early feedback on whether your planned manuscript fits the scope of this Research Topic, please write to us at plantscience@frontiersin.org before submitting. A quick check at this stage helps confirm your work is a strong fit and can save you time later in the process.To take part, click “Participate in this topic” at the top right of this page, then confirm through the “Participate” link in the follow-up email. This lets us know to expect your contribution, keeps you informed of key deadlines, and gives you access to resources that guide you through submission. We look forward to your participation and to building a vibrant, cohesive Research Topic together. Topic editors: Pan Gao, Shijie Tian, Jinshan Yan, Monica Herrero Huerta Source: https://www.frontiersin.org/research-topics/81899/multimodal-ai-phenotyping-for-sustainable-crop-production-structure-to-function-trait-quantification-from-non-destructive-2d-and-3d-sensing