Generative Adversarial Networks (GAN) involve the competition between a generator and a discriminator. The large variance of training data can bring difficulty to GAN resulting in mode collapse, training instability and low quality. As the distribution of generated samples evolves, the discriminator will have different levels of confidence for its predictions. To mitigate these problems, we introduce confidence-aware discrimination (CAD) to guide the training and sampling of GANs. To adapt confidence estimation for GANs, we design a new architecture based on StyleGAN2, and propose a confidence-aware adversarial loss. Extensive experiments are conducted on face, scene and object generation benchmarks. In the training stage, CAD can progressively learn the training data with the guide of confidence estimation and lead to a better convergence. In the inference stage, the confidence score can provide a new dimension of metric to assess the quality of generated samples and can further improve the performance by resampling and finetuning.

Deep image restoration models aim to learn a mapping from degraded image space to natural image space. However, they face several critical challenges: removing degradation, generating realistic details, and ensuring pixel-level consistency. Over time, three major classes of methods have emerged, including MSE-based, GAN-based, and diffusion-based methods. However, they fail to achieve a good balance between restoration quality, fidelity, and speed. We propose a novel method, HYPIR, to address these challenges. Our solution pipeline is straightforward: it involves initializing the image restoration model with a pre-trained diffusion model and then fine-tuning it with adversarial training. This approach does not rely on diffusion loss, iterative sampling, or additional adapters. We theoretically demonstrate that initializing adversarial training from a pre-trained diffusion model positions the initial restoration model very close to the natural image distribution. Consequently, this initialization improves numerical stability, avoids mode collapse, and substantially accelerates the convergence of adversarial training. Moreover, HYPIR inherits the capabilities of diffusion models with rich user control, enabling text-guided restoration and adjustable texture richness. Requiring only a single forward pass, it achieves faster convergence and inference speed than diffusion-based methods. Extensive experiments show that HYPIR outperforms previous state-of-the-art methods, achieving efficient and high-quality image restoration.

Data uncertainty is commonly observed in the images for face recognition (FR). However, deep learning algorithms often make predictions with high confidence even for uncertain or irrelevant inputs. Intuitively, FR algorithms can benefit from both the estimation of uncertainty and the detection of out-of-distribution (OOD) samples. Taking a probabilistic view of the current classification model, the temperature scalar is exactly the scale of uncertainty noise implicitly added in the softmax function. Meanwhile, the uncertainty of images in a dataset should follow a prior distribution. Based on the observation, a unified framework for uncertainty modeling and FR, Random Temperature Scaling (RTS), is proposed to learn a reliable FR algorithm. The benefits of RTS are two-fold. (1) In the training phase, it can adjust the learning strength of clean and noisy samples for stability and accuracy. (2) In the test phase, it can provide a score of confidence to detect uncertain, low-quality and even OOD samples, without training on extra labels. Extensive experiments on FR benchmarks demonstrate that the magnitude of variance in RTS, which serves as an OOD detection metric, is closely related to the uncertainty of the input image. RTS can achieve top performance on both the FR and OOD detection tasks. Moreover, the model trained with RTS can perform robustly on datasets with noise. The proposed module is light-weight and only adds negligible computation cost to the model.

Face restoration models are usually trained on synthetic degraded data to output an image that matches the clean version of itself. Most previous methods use a single model to deal with all the degradation levels, resulting in a domain generalization problem. We explore the value of degradation information and propose a Degradation Conditioned GAN (DeCGAN). The architecture consists of modulated convolution, bias, and fusion modules, inspired by deblurring, denoising and super-resolution. The whole network can be modulated by the degradation levels to achieve delicate and precise restoration effects. Experiments are conducted on conventional and modulated face restoration tasks. DeCGAN can achieve more faithful restoration and better metrics (FID, LPIPS, etc.) than previous methods do. Moreover, our model performs well on real-world low-quality face images.

Face image super resolution (face hallucination) usu-ally relies on facial priors to restore realistic details and preserve identity information. Recent advances can achieve impressive results with the help of GAN prior. They ei-ther design complicated modules to modify the fixed GAN prior or adopt complex training strategies to finetune the generator. In this work, we propose a generative and controllable face SR framework, called GCFSR, which can re-construct images with faithful identity information without any additional priors. Generally, GCFSR has an encoder-generator architecture. Two modules called style modu-lation and feature modulation are designed for the multi-factor SR task. The style modulation aims to generate real-istic face details and the feature modulation dynamically fuses the multi-level encoded features and the generated ones conditioned on the upscaling factor. The simple and elegant architecture can be trained from scratch in an end-to-end manner. For small upscaling factors (≤8), GCFSR can produce surprisingly good results with only adversar-ialloss. After adding L1 and perceptual losses, GCFSR can outperform state-of-the-art methods for large upscalingfac-tors (16, 32, 64). During the test phase, we can modulate the generative strength via feature modulation by changing the conditional upscaling factor continuously to achieve various generative effects. Code is available at https://github.com/hejingwenhejingwen/GCFSR

Texture synthesis using deep neural networks can generate high quality and diversified textures. However, it usually requires a heavy optimization process. The following works accelerate the process by using feed-forward networks, but at the cost of scalability. diversity or quality. We propose a new efficient method that aims to simulate the optimization process while retains most of the properties. Our method takes a noise image and the gradients from a descriptor network as inputs, and synthesize a refined image with respect to the target image. The proposed method can synthesize images with better quality and diversity than the other fast synthesis methods do. Moreover, our method trained on a large scale dataset can generalize to synthesize unseen textures.