TaxonRL: Reinforcement Learning with Intermediate Rewards for Interpretable Fine-Grained Visual Reasoning

TL;DR

TaxonRL, a reinforcement learning approach using Group Relative Policy Optimization with intermediate rewards that decomposes the reasoning process into hierarchical taxonomic predictions, is introduced, establishing that enforcing structured, hierarchical reasoning provides a powerful and transferable framework for fine-grained visual discrimination.

Abstract

Traditional vision-language models struggle with contrastive fine-grained taxonomic reasoning, particularly when distinguishing between visually similar species within the same genus or family. We introduce TaxonRL, a reinforcement learning approach using Group Relative Policy Optimization with intermediate rewards that decomposes the reasoning process into hierarchical taxonomic predictions. Our method incentivizes models to explicitly reason about species-level, genus-level, and family-level features before making final classifications. This structured approach is designed not only to boost accuracy but also to yield a transparent, verifiable decision-making process. On the challenging Birds-to-Words dataset, TaxonRL achieves 91.7\% average accuracy, exceeding human performance (77.3\%) while generating interpretable reasoning traces. We demonstrate strong cross-domain generalization, showing substantial gains in primate and marine species verification. Our results establish that enforcing structured, hierarchical reasoning provides a powerful and transferable framework for fine-grained visual discrimination.

Figures (1)

• Figure 1: Hierarchical Reasoning Pipeline. An example from the Birds-to-Words dataset. The model performs a systematic taxonomic verification (Order $\rightarrow$ Family $\rightarrow$ Genus), grounding each step in visual features to produce an interpretable chain-of-thought and final confidence score.