Semi-supervised class-conditional image synthesis is typically performed by inferring and injecting class labels into a conditional Generative Adversarial Network (GAN). The supervision in the form of class identity may be inadequate to model classes with diverse visual appearances. In this paper, we propose a Learnable Cluster Prompt-based GAN (LCP-GAN) to capture class-wise characteristics and intra-class variation factors with a broader source of supervision. To exploit partially labeled data, we perform soft partitioning on each class, and explore the possibility of associating intra-class clusters with learnable visual concepts in the feature space of a pre-trained language-vision model, e.g., CLIP. For class-conditional image generation, we design a cluster-conditional generator by injecting a combination of intra-class cluster label embeddings, and further incorporate a real-fake classification head on top of CLIP to distinguish real instances from the synthesized ones, conditioned on the learnable cluster prompts. This significantly strengthens the generator with more semantic language supervision. LCP-GAN not only possesses superior generation capability but also matches the performance of the fully supervised version of the base models: BigGAN and StyleGAN2-ADA, on multiple standard benchmarks.

Generative Adversarial Network (GAN)-based models have greatly facilitated image synthesis. However, the model performance may be degraded when applied to finegrained data, due to limited training samples and subtle distinction among categories. Different from generic GAN-s, we address the issue from a new perspective of discovering and utilizing the underlying structure of real data to explicitly regularize the spatial organization of latent space. To reduce the dependence of generative models on labeled data, we propose a semi-supervised hyper-spherical GAN for class-conditional fine-grained image generation, and our model is referred to as SphericGAN. By projecting random vectors drawn from a prior distribution onto a hyper-sphere, we can model more complex distributions, while at the same time the similarity between the resulting latent vectors depends only on the angle, but not on their magnitudes. On the other hand, we also incorporate a mapping network to map real images onto the hyper-sphere, and match latent vectors with the underlying structure of real data via real-fake cluster alignment. As a result, we obtain a spatially organized latent space, which is useful for capturing class-independent variation factors. The experi-mental results suggest that our SphericGAN achieves state-of-the-art performance in synthesizing high-fidelity images with precise class semantics.

Generative Adversarial Network (GAN)-based models have made remarkable progress in high-fidelity image synthesis. However, the performance of class-conditional image synthesis may significantly deteriorate for the case where a limited number of labeled training samples are available. To reduce the dependence on labeled data, we propose a semi-supervised GAN with Extended Distribution Matching, and our model is referred to as EDM-GAN. To prevent a class-conditional discriminator from overfitting the limited labeled data, we perform a transformation of random regional replacement on both real and synthesized samples. By matching the extended distributions, the discriminator is encouraged to focus more on the spatial regions that contain certain objects, while at the same time a class-conditional generator is induced to capture precise class semantics. The adversarial training process can be effectively stabilized and converges to a better solution. Our experimental results on multiple standard benchmarks demonstrate consistent performance gains in synthesis quality and class-semantic accuracy.

Semi-supervised generative learning (SSGL) makes use of unlabeled data to achieve a trade-off between the data collection/annotation effort and generation performance, when adequate labeled data are not available. Learning precise class semantics is crucial for class-conditional image synthesis with limited supervision. Toward this end, we propose a semi-supervised Generative Adversarial Network with a Mask-Embedded Discriminator, which is referred to as MED-GAN. By incorporating a mask embedding module, the discriminator features are associated with spatial information, such that the focus of the discriminator can be limited in the specified regions when distinguishing between real and synthesized images. A generator is enforced to synthesize the instances holding more precise class semantics in order to deceive the enhanced discriminator. Also benefiting from mask embedding, region-based semantic regularization is imposed on the discriminator feature space, and the degree of separation between real and fake classes and among object categories can thus be increased. This eventually improves class-conditional distribution matching between real and synthesized data. In the experiments, the superior performance of MED-GAN demonstrates the effectiveness of mask embedding and associated regularizers in facilitating SSGL.