Generative Adversarial Networks with Limited Data: A Survey and Benchmarking
Omar De Mitri, Ruyu Wang, Marco F. Huber
TL;DR
This survey analyzes Generative Adversarial Networks in data-scarce regimes, surveying architectures, augmentation strategies, few-shot methods, and GAN inversion techniques. It benchmarks state-of-the-art models across image synthesis, semantic synthesis, and image-to-image translation on limited data, revealing that semantic conditioning provides the strongest robustness while unconditional models degrade rapidly. The study highlights practical data-efficiency strategies (ADA, DiffAugment, GAN inversion-based editing) and discusses remaining challenges, such as mode collapse and artifact generation, with implications for real-world deployment in domains with scarce labeled data. It also outlines promising directions toward data-efficient, controllable generation and scalable labeling through synthetic data pipelines and latent-space manipulation.
Abstract
Generative Adversarial Networks (GANs) have shown impressive results in various image synthesis tasks. Vast studies have demonstrated that GANs are more powerful in feature and expression learning compared to other generative models and their latent space encodes rich semantic information. However, the tremendous performance of GANs heavily relies on the access to large-scale training data and deteriorates rapidly when the amount of data is limited. This paper aims to provide an overview of GANs, its variants and applications in various vision tasks, focusing on addressing the limited data issue. We analyze state-of-the-art GANs in limited data regime with designed experiments, along with presenting various methods attempt to tackle this problem from different perspectives. Finally, we further elaborate on remaining challenges and trends for future research.