TaxaDiffusion: Progressively Trained Diffusion Model for Fine-Grained Species Generation

TL;DR

TaxaDiffusion introduces a hierarchical training framework for diffusion models that leverages taxonomic relationships to generate high-fidelity, fine-grained animal images, especially effective with limited data.

Contribution

It proposes a progressive, taxonomy-informed training approach for diffusion models that improves fine-grained animal image generation by capturing hierarchical morphological features.

Findings

Outperforms existing methods in fidelity of animal image generation

Effective with limited training samples per species

Demonstrates superior results on three fine-grained datasets

Abstract

We propose TaxaDiffusion, a taxonomy-informed training framework for diffusion models to generate fine-grained animal images with high morphological and identity accuracy. Unlike standard approaches that treat each species as an independent category, TaxaDiffusion incorporates domain knowledge that many species exhibit strong visual similarities, with distinctions often residing in subtle variations of shape, pattern, and color. To exploit these relationships, TaxaDiffusion progressively trains conditioned diffusion models across different taxonomic levels -- starting from broad classifications such as Class and Order, refining through Family and Genus, and ultimately distinguishing at the Species level. This hierarchical learning strategy first captures coarse-grained morphological traits shared by species with common ancestors, facilitating knowledge transfer before refining fine-grained differences for species-level distinction. As a result, TaxaDiffusion enables accurate generation even with limited training samples per species. Extensive experiments on three fine-grained animal datasets demonstrate that outperforms existing approaches, achieving superior fidelity in fine-grained animal image generation. Project page: https://amink8.github.io/TaxaDiffusion/