TreeFormer: Single-view Plant Skeleton Estimation via Tree-constrained Graph Generation
This work addresses a domain-specific problem in smart agriculture and plant science by providing a more accurate tool for plant skeleton estimation, though it is incremental as it builds on existing graph generation methods with added constraints.
The paper tackles the problem of estimating plant skeletal structures from single-view images, which is challenging due to the need for arbitrary tree graphs, and presents TreeFormer, a method that combines learning-based graph generation with graph algorithms to enforce tree constraints, achieving accurate results on synthetic tree patterns, real botanical roots, and grapevine branches.
Accurate estimation of plant skeletal structure (e.g., branching structure) 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. While recent graph generation methods successfully infer thin structures from images, it is challenging to constrain the output graph strictly to a tree structure. To this problem, we present TreeFormer, a plant skeleton estimator via tree-constrained graph generation. Our approach combines learning-based graph generation with traditional graph algorithms to impose the constraints during the training loop. Specifically, our method projects an unconstrained graph onto a minimum spanning tree (MST) during the training loop and incorporates this prior knowledge into the gradient descent optimization by suppressing unwanted feature values. Experiments show that our method accurately estimates target plant skeletal structures for multiple domains: Synthetic tree patterns, real botanical roots, and grapevine branches. Our implementations are available at https://github.com/huntorochi/TreeFormer/.