Dual-Margin Embedding for Fine-Grained Long-Tailed Plant Taxonomy

TL;DR

TaxoNet is a novel embedding framework with a dual-margin objective that enhances fine-grained, long-tailed plant taxonomy classification, especially under open-world and domain shift conditions.

Contribution

It introduces a theoretically grounded dual-margin loss to improve class discrimination and rare-class representation in hierarchical, long-tailed plant taxonomy tasks.

Findings

TaxoNet outperforms strong baselines on diverse plant datasets.

It demonstrates robustness under open-world recognition challenges.

The dual-margin approach effectively reshapes decision boundaries for imbalanced classes.

Abstract

Taxonomic classification of ecological families, genera, and species underpins biodiversity monitoring and conservation. Existing computer vision methods typically address fine-grained recognition and long-tailed learning in isolation. However, additional challenges such as spatiotemporal domain shift, hierarchical taxonomic structure, and previously unseen taxa often co-occur in real-world deployment, leading to brittle performance under open-world conditions. We propose TaxoNet, an embedding learning framework with a theoretically grounded dual-margin objective that reshapes class decision boundaries under class imbalance to improve fine-grained discrimination while strengthening rare-class representation geometry. We evaluate TaxoNet in open-world settings that capture co-occurring recognition challenges. Leveraging diverse plant datasets, including Google Auto-Arborist (urban tree imagery), iNaturalist (Plantae observations across heterogeneous ecosystems), and NAFlora-Mini (herbarium collections), we demonstrate that TaxoNet consistently outperforms strong baselines, including multimodal foundation models.