Image generation diversity issues and how to tame them
OA Location
Author(s)
Dombrowski, Mischa
Zhang, Weitong
Cechnicka, Sarah
Reynaud, Hadrien
Kainz, Bernhard
Type
Conference Paper
Abstract
Generative methods have reached a level of quality that is almost indistinguishable from real data. However, while individual samples may appear unique, generative models often exhibit limitations in covering the full data distribution. Unlike quality issues, diversity problems within generative models are not easily detected by simply observing single images or generated datasets, which means we need a specific measure to assess the diversity of these models. In this paper, we draw attention to the current lack of diversity in generative models and the inability of common metrics to measure this. We achieve this by framing diversity as an image retrieval problem, where we measure how many real images can be retrieved using synthetic data as queries. This yields the Image Retrieval Score (IRS), an interpretable, hyperparameter-free metric that quantifies the diversity of a generative model’s output. IRS requires only a subset of synthetic samples and provides a statistical measure of confidence. Our experiments indicate that current feature extractors commonly used in generative model assessment are inadequate for evaluating diversity effectively. Consequently, we perform an extensive search for the best feature extractors to assess diversity. Evaluation reveals that current diffusion models converge to limited subsets of the real distribution, with no current state-of-the-art models superpassing 77% of the diversity of the training data. To address this limitation, we introduce Diversity-Aware Diffusion Models (DiADM), a novel approach that improves diversity of unconditional diffusion models without loss of image quality. We do this by disentangling diversity from image quality by using a diversity aware module that uses pseudo-unconditional features as input. We provide a Python package offering unified feature extraction and metric computation to further facilitate the evaluation of generative models https://github.com/MischaD/beyondfid.
Date Issued
2025-08-13
Date Acceptance
2025-06-01
Citation
2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp.3029-3039
ISBN
979-8-3315-4365-5
ISSN
1063-6919
Publisher
IEEE Computer Society
Start Page
3029
End Page
3039
Journal / Book Title
2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Copyright Statement
© 2025 IEEE.
Source
CVPR 2025
Subjects
Computer Science
Computer Science, Artificial Intelligence
Computer Science, Interdisciplinary Applications
Computer Science, Theory & Methods
Science & Technology
Technology
Publication Status
Published
Start Date
2025-06-10
Finish Date
2025-06-17
Coverage Spatial
Nashville, TN, USA