Research

Paper

TESTING March 04, 2026

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

Authors

Maximilian von Klinski, Maximilian Schall

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.

Metadata

arXiv ID: 2603.04380
Provider: ARXIV
Primary Category: cs.CV
Published: 2026-03-04
Fetched: 2026-03-05 06:06

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.04380v1</id>\n    <title>TaxonRL: Reinforcement Learning with Intermediate Rewards for Interpretable Fine-Grained Visual Reasoning</title>\n    <updated>2026-03-04T18:45:35Z</updated>\n    <link href='https://arxiv.org/abs/2603.04380v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.04380v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>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.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CV'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.CL'/>\n    <published>2026-03-04T18:45:35Z</published>\n    <arxiv:comment>Accepted at WACV 2026</arxiv:comment>\n    <arxiv:primary_category term='cs.CV'/>\n    <author>\n      <name>Maximilian von Klinski</name>\n    </author>\n    <author>\n      <name>Maximilian Schall</name>\n    </author>\n  </entry>"
}