Structure-aware Riemannian Growth Fields for 4D Plant Modeling
This work addresses the challenge of modeling rapid organ emergence, drastic topological changes, and self-occlusion in 4D plant growth under sparse observations and large temporal gaps. The authors propose to formulate plant morphogenesis as a continuous process on a structure-aware Riemannian growth field, where symbolically guided geodesic flows jointly optimize geometric deformation and topological evolution to establish stable spatiotemporal correspondences across time points. By integrating structure-aware Riemannian geometry with symbolic growth rules—a first in the field—the method preserves structural consistency despite topological alterations. The study also introduces the first high-annotation-density dataset spanning ten days across two plant species. Experiments demonstrate that the proposed approach significantly outperforms existing methods in both geometric accuracy and organ correspondence consistency, enabling precise long-term tracking of individual organ growth trajectories even from sparse observations.