PlantPose: Universal Plant Skeleton Estimation via Tree-constrained Graph Generation

2026-05-17

AI Summary

Existing plant skeleton estimation methods are constrained by fixed topological assumptions, limiting their ability to handle arbitrary tree-like structures. This work proposes PlantPose, the first approach to embed explicit tree-structure constraints into an end-to-end graph generation pipeline, integrating learning-driven graph generation with graph algorithms to enforce topological consistency during training. The method establishes a cross-domain generalizable framework for plant skeleton estimation and introduces a multi-source heterogeneous dataset encompassing real, synthetic, and abstract-style data to enhance model robustness and generalization. Experiments demonstrate that PlantPose achieves robust, accurate, and topologically consistent skeleton estimation across diverse plant image domains, including unseen out-of-distribution scenarios.

Abstract

Accurate estimation of plant skeletal structures (e.g., branching structures) from images is essential for smart agriculture and plant science. Unlike human skeletons with fixed topology, plant skeleton estimation presents a unique challenge, i.e., estimating arbitrary tree graphs from images. To address this problem, we introduce PlantPose, a universal plant skeleton estimator via tree-constrained graph generation. PlantPose combines learning-based graph generation with traditional graph algorithms to enforce tree constraints during the training loop. To enhance the model's generalization capability, we curate a large and diverse dataset comprising real-world and synthetic plant images, along with simplified representations (e.g., sketches and abstract drawings). This dataset enables the generalized model to adapt to diverse input styles and categories of plant images while preserving topological consistency. Our approach demonstrates robust and accurate plant skeleton estimation across multiple domains, including previously unseen out-of-domain scenarios. Further analyses highlight the method's strengths and limitations in handling complex, heterogeneous data distributions. All implementations and datasets are available at https://github.com/huntorochi/PlantPose/.

Problem

• plant skeleton estimation
• tree graph
• arbitrary topology
• smart agriculture
• plant science

Innovation

• tree-constrained graph generation
• plant skeleton estimation
• topological consistency
• cross-domain generalization
• graph neural networks