Peyman Milanfar

Zhengzhong Tu, Hossein Talebi, Han Zhang et al.

Transformers have recently gained significant attention in the computer vision community. However, the lack of scalability of self-attention mechanisms with respect to image size has limited their wide adoption in state-of-the-art vision backbones. In this paper we introduce an efficient and scalable attention model we call multi-axis attention, which consists of two aspects: blocked local and dilated global attention. These design choices allow global-local spatial interactions on arbitrary input resolutions with only linear complexity. We also present a new architectural element by effectively blending our proposed attention model with convolutions, and accordingly propose a simple hierarchical vision backbone, dubbed MaxViT, by simply repeating the basic building block over multiple stages. Notably, MaxViT is able to ''see'' globally throughout the entire network, even in earlier, high-resolution stages. We demonstrate the effectiveness of our model on a broad spectrum of vision tasks. On image classification, MaxViT achieves state-of-the-art performance under various settings: without extra data, MaxViT attains 86.5% ImageNet-1K top-1 accuracy; with ImageNet-21K pre-training, our model achieves 88.7% top-1 accuracy. For downstream tasks, MaxViT as a backbone delivers favorable performance on object detection as well as visual aesthetic assessment. We also show that our proposed model expresses strong generative modeling capability on ImageNet, demonstrating the superior potential of MaxViT blocks as a universal vision module. The source code and trained models will be available at https://github.com/google-research/maxvit.

Ligong Han, Yinxiao Li, Han Zhang et al.

Diffusion models have achieved remarkable success in text-to-image generation, enabling the creation of high-quality images from text prompts or other modalities. However, existing methods for customizing these models are limited by handling multiple personalized subjects and the risk of overfitting. Moreover, their large number of parameters is inefficient for model storage. In this paper, we propose a novel approach to address these limitations in existing text-to-image diffusion models for personalization. Our method involves fine-tuning the singular values of the weight matrices, leading to a compact and efficient parameter space that reduces the risk of overfitting and language drifting. We also propose a Cut-Mix-Unmix data-augmentation technique to enhance the quality of multi-subject image generation and a simple text-based image editing framework. Our proposed SVDiff method has a significantly smaller model size compared to existing methods (approximately 2,200 times fewer parameters compared with vanilla DreamBooth), making it more practical for real-world applications.

Mauricio Delbracio, Peyman Milanfar

Inversion by Direct Iteration (InDI) is a new formulation for supervised image restoration that avoids the so-called "regression to the mean" effect and produces more realistic and detailed images than existing regression-based methods. It does this by gradually improving image quality in small steps, similar to generative denoising diffusion models. Image restoration is an ill-posed problem where multiple high-quality images are plausible reconstructions of a given low-quality input. Therefore, the outcome of a single step regression model is typically an aggregate of all possible explanations, therefore lacking details and realism. The main advantage of InDI is that it does not try to predict the clean target image in a single step but instead gradually improves the image in small steps, resulting in better perceptual quality. While generative denoising diffusion models also work in small steps, our formulation is distinct in that it does not require knowledge of any analytic form of the degradation process. Instead, we directly learn an iterative restoration process from low-quality and high-quality paired examples. InDI can be applied to virtually any image degradation, given paired training data. In conditional denoising diffusion image restoration the denoising network generates the restored image by repeatedly denoising an initial image of pure noise, conditioned on the degraded input. Contrary to conditional denoising formulations, InDI directly proceeds by iteratively restoring the input low-quality image, producing high-quality results on a variety of image restoration tasks, including motion and out-of-focus deblurring, super-resolution, compression artifact removal, and denoising.

Mengwei Ren, Mauricio Delbracio, Hossein Talebi et al.

Diffusion Probabilistic Models (DPMs) have recently been employed for image deblurring, formulated as an image-conditioned generation process that maps Gaussian noise to the high-quality image, conditioned on the blurry input. Image-conditioned DPMs (icDPMs) have shown more realistic results than regression-based methods when trained on pairwise in-domain data. However, their robustness in restoring images is unclear when presented with out-of-domain images as they do not impose specific degradation models or intermediate constraints. To this end, we introduce a simple yet effective multiscale structure guidance as an implicit bias that informs the icDPM about the coarse structure of the sharp image at the intermediate layers. This guided formulation leads to a significant improvement of the deblurring results, particularly on unseen domain. The guidance is extracted from the latent space of a regression network trained to predict the clean-sharp target at multiple lower resolutions, thus maintaining the most salient sharp structures. With both the blurry input and multiscale guidance, the icDPM model can better understand the blur and recover the clean image. We evaluate a single-dataset trained model on diverse datasets and demonstrate more robust deblurring results with fewer artifacts on unseen data. Our method outperforms existing baselines, achieving state-of-the-art perceptual quality while keeping competitive distortion metrics.

Giannis Daras, Hyungjin Chung, Chieh-Hsin Lai et al.

Diffusion models have become increasingly popular for generative modeling due to their ability to generate high-quality samples. This has unlocked exciting new possibilities for solving inverse problems, especially in image restoration and reconstruction, by treating diffusion models as unsupervised priors. This survey provides a comprehensive overview of methods that utilize pre-trained diffusion models to solve inverse problems without requiring further training. We introduce taxonomies to categorize these methods based on both the problems they address and the techniques they employ. We analyze the connections between different approaches, offering insights into their practical implementation and highlighting important considerations. We further discuss specific challenges and potential solutions associated with using latent diffusion models for inverse problems. This work aims to be a valuable resource for those interested in learning about the intersection of diffusion models and inverse problems.

Giannis Daras, Mauricio Delbracio, Hossein Talebi et al.

We define a broader family of corruption processes that generalizes previously known diffusion models. To reverse these general diffusions, we propose a new objective called Soft Score Matching that provably learns the score function for any linear corruption process and yields state of the art results for CelebA. Soft Score Matching incorporates the degradation process in the network. Our new loss trains the model to predict a clean image, \textit{that after corruption}, matches the diffused observation. We show that our objective learns the gradient of the likelihood under suitable regularity conditions for a family of corruption processes. We further develop a principled way to select the corruption levels for general diffusion processes and a novel sampling method that we call Momentum Sampler. We show experimentally that our framework works for general linear corruption processes, such as Gaussian blur and masking. We achieve state-of-the-art FID score $1.85$ on CelebA-64, outperforming all previous linear diffusion models. We also show significant computational benefits compared to vanilla denoising diffusion.

Junjie Ke, Keren Ye, Jiahui Yu et al.

Assessing the aesthetics of an image is challenging, as it is influenced by multiple factors including composition, color, style, and high-level semantics. Existing image aesthetic assessment (IAA) methods primarily rely on human-labeled rating scores, which oversimplify the visual aesthetic information that humans perceive. Conversely, user comments offer more comprehensive information and are a more natural way to express human opinions and preferences regarding image aesthetics. In light of this, we propose learning image aesthetics from user comments, and exploring vision-language pretraining methods to learn multimodal aesthetic representations. Specifically, we pretrain an image-text encoder-decoder model with image-comment pairs, using contrastive and generative objectives to learn rich and generic aesthetic semantics without human labels. To efficiently adapt the pretrained model for downstream IAA tasks, we further propose a lightweight rank-based adapter that employs text as an anchor to learn the aesthetic ranking concept. Our results show that our pretrained aesthetic vision-language model outperforms prior works on image aesthetic captioning over the AVA-Captions dataset, and it has powerful zero-shot capability for aesthetic tasks such as zero-shot style classification and zero-shot IAA, surpassing many supervised baselines. With only minimal finetuning parameters using the proposed adapter module, our model achieves state-of-the-art IAA performance over the AVA dataset.

Hyungjin Chung, Jong Chul Ye, Peyman Milanfar et al.

We propose a new method for solving imaging inverse problems using text-to-image latent diffusion models as general priors. Existing methods using latent diffusion models for inverse problems typically rely on simple null text prompts, which can lead to suboptimal performance. To address this limitation, we introduce a method for prompt tuning, which jointly optimizes the text embedding on-the-fly while running the reverse diffusion process. This allows us to generate images that are more faithful to the diffusion prior. In addition, we propose a method to keep the evolution of latent variables within the range space of the encoder, by projection. This helps to reduce image artifacts, a major problem when using latent diffusion models instead of pixel-based diffusion models. Our combined method, called P2L, outperforms both image- and latent-diffusion model-based inverse problem solvers on a variety of tasks, such as super-resolution, deblurring, and inpainting.

Peyman Milanfar, Mauricio Delbracio

Denoising, the process of reducing random fluctuations in a signal to emphasize essential patterns, has been a fundamental problem of interest since the dawn of modern scientific inquiry. Recent denoising techniques, particularly in imaging, have achieved remarkable success, nearing theoretical limits by some measures. Yet, despite tens of thousands of research papers, the wide-ranging applications of denoising beyond noise removal have not been fully recognized. This is partly due to the vast and diverse literature, making a clear overview challenging. This paper aims to address this gap. We present a clarifying perspective on denoisers, their structure, and desired properties. We emphasize the increasing importance of denoising and showcase its evolution into an essential building block for complex tasks in imaging, inverse problems, and machine learning. Despite its long history, the community continues to uncover unexpected and groundbreaking uses for denoising, further solidifying its place as a cornerstone of scientific and engineering practice.

Kangfu Mei, Mauricio Delbracio, Hossein Talebi et al.

Large generative diffusion models have revolutionized text-to-image generation and offer immense potential for conditional generation tasks such as image enhancement, restoration, editing, and compositing. However, their widespread adoption is hindered by the high computational cost, which limits their real-time application. To address this challenge, we introduce a novel method dubbed CoDi, that adapts a pre-trained latent diffusion model to accept additional image conditioning inputs while significantly reducing the sampling steps required to achieve high-quality results. Our method can leverage architectures such as ControlNet to incorporate conditioning inputs without compromising the model's prior knowledge gained during large scale pre-training. Additionally, a conditional consistency loss enforces consistent predictions across diffusion steps, effectively compelling the model to generate high-quality images with conditions in a few steps. Our conditional-task learning and distillation approach outperforms previous distillation methods, achieving a new state-of-the-art in producing high-quality images with very few steps (e.g., 1-4) across multiple tasks, including super-resolution, text-guided image editing, and depth-to-image generation.

Junjie Ke, Tianhao Zhang, Yilin Wang et al.

No-reference video quality assessment (NR-VQA) for user generated content (UGC) is crucial for understanding and improving visual experience. Unlike video recognition tasks, VQA tasks are sensitive to changes in input resolution. Since large amounts of UGC videos nowadays are 720p or above, the fixed and relatively small input used in conventional NR-VQA methods results in missing high-frequency details for many videos. In this paper, we propose a novel Transformer-based NR-VQA framework that preserves the high-resolution quality information. With the multi-resolution input representation and a novel multi-resolution patch sampling mechanism, our method enables a comprehensive view of both the global video composition and local high-resolution details. The proposed approach can effectively aggregate quality information across different granularities in spatial and temporal dimensions, making the model robust to input resolution variations. Our method achieves state-of-the-art performance on large-scale UGC VQA datasets LSVQ and LSVQ-1080p, and on KoNViD-1k and LIVE-VQC without fine-tuning.

Zhengzhong Tu, Peyman Milanfar, Hossein Talebi

Image resizing operation is a fundamental preprocessing module in modern computer vision. Throughout the deep learning revolution, researchers have overlooked the potential of alternative resizing methods beyond the commonly used resizers that are readily available, such as nearest-neighbors, bilinear, and bicubic. The key question of our interest is whether the front-end resizer affects the performance of deep vision models? In this paper, we present an extremely lightweight multilayer Laplacian resizer with only a handful of trainable parameters, dubbed MULLER resizer. MULLER has a bandpass nature in that it learns to boost details in certain frequency subbands that benefit the downstream recognition models. We show that MULLER can be easily plugged into various training pipelines, and it effectively boosts the performance of the underlying vision task with little to no extra cost. Specifically, we select a state-of-the-art vision Transformer, MaxViT, as the baseline, and show that, if trained with MULLER, MaxViT gains up to 0.6% top-1 accuracy, and meanwhile enjoys 36% inference cost saving to achieve similar top-1 accuracy on ImageNet-1k, as compared to the standard training scheme. Notably, MULLER's performance also scales with model size and training data size such as ImageNet-21k and JFT, and it is widely applicable to multiple vision tasks, including image classification, object detection and segmentation, as well as image quality assessment.

William Yicheng Zhu, Keren Ye, Junjie Ke et al.

Recognizing and disentangling visual attributes from objects is a foundation to many computer vision applications. While large vision language representations like CLIP had largely resolved the task of zero-shot object recognition, zero-shot visual attribute recognition remains a challenge because CLIP's contrastively-learned vision-language representation cannot effectively capture object-attribute dependencies. In this paper, we target this weakness and propose a sentence generation-based retrieval formulation for attribute recognition that is novel in 1) explicitly modeling a to-be-measured and retrieved object-attribute relation as a conditional probability graph, which converts the recognition problem into a dependency-sensitive language-modeling problem, and 2) applying a large pretrained Vision-Language Model (VLM) on this reformulation and naturally distilling its knowledge of image-object-attribute relations to use towards attribute recognition. Specifically, for each attribute to be recognized on an image, we measure the visual-conditioned probability of generating a short sentence encoding the attribute's relation to objects on the image. Unlike contrastive retrieval, which measures likelihood by globally aligning elements of the sentence to the image, generative retrieval is sensitive to the order and dependency of objects and attributes in the sentence. We demonstrate through experiments that generative retrieval consistently outperforms contrastive retrieval on two visual reasoning datasets, Visual Attribute in the Wild (VAW), and our newly-proposed Visual Genome Attribute Ranking (VGARank).

Zhengzhong Tu, Hossein Talebi, Han Zhang et al.

Recent progress on Transformers and multi-layer perceptron (MLP) models provide new network architectural designs for computer vision tasks. Although these models proved to be effective in many vision tasks such as image recognition, there remain challenges in adapting them for low-level vision. The inflexibility to support high-resolution images and limitations of local attention are perhaps the main bottlenecks. In this work, we present a multi-axis MLP based architecture called MAXIM, that can serve as an efficient and flexible general-purpose vision backbone for image processing tasks. MAXIM uses a UNet-shaped hierarchical structure and supports long-range interactions enabled by spatially-gated MLPs. Specifically, MAXIM contains two MLP-based building blocks: a multi-axis gated MLP that allows for efficient and scalable spatial mixing of local and global visual cues, and a cross-gating block, an alternative to cross-attention, which accounts for cross-feature conditioning. Both these modules are exclusively based on MLPs, but also benefit from being both global and `fully-convolutional', two properties that are desirable for image processing. Our extensive experimental results show that the proposed MAXIM model achieves state-of-the-art performance on more than ten benchmarks across a range of image processing tasks, including denoising, deblurring, deraining, dehazing, and enhancement while requiring fewer or comparable numbers of parameters and FLOPs than competitive models. The source code and trained models will be available at \url{https://github.com/google-research/maxim}.

Yinxiao Li, Pengchong Jin, Feng Yang et al.

Most video super-resolution methods focus on restoring high-resolution video frames from low-resolution videos without taking into account compression. However, most videos on the web or mobile devices are compressed, and the compression can be severe when the bandwidth is limited. In this paper, we propose a new compression-informed video super-resolution model to restore high-resolution content without introducing artifacts caused by compression. The proposed model consists of three modules for video super-resolution: bi-directional recurrent warping, detail-preserving flow estimation, and Laplacian enhancement. All these three modules are used to deal with compression properties such as the location of the intra-frames in the input and smoothness in the output frames. For thorough performance evaluation, we conducted extensive experiments on standard datasets with a wide range of compression rates, covering many real video use cases. We showed that our method not only recovers high-resolution content on uncompressed frames from the widely-used benchmark datasets, but also achieves state-of-the-art performance in super-resolving compressed videos based on numerous quantitative metrics. We also evaluated the proposed method by simulating streaming from YouTube to demonstrate its effectiveness and robustness. The source codes and trained models are available at https://github.com/google-research/google-research/tree/master/comisr.

Kangfu Mei, Zhengzhong Tu, Mauricio Delbracio et al.

We study the scaling properties of latent diffusion models (LDMs) with an emphasis on their sampling efficiency. While improved network architecture and inference algorithms have shown to effectively boost sampling efficiency of diffusion models, the role of model size -- a critical determinant of sampling efficiency -- has not been thoroughly examined. Through empirical analysis of established text-to-image diffusion models, we conduct an in-depth investigation into how model size influences sampling efficiency across varying sampling steps. Our findings unveil a surprising trend: when operating under a given inference budget, smaller models frequently outperform their larger equivalents in generating high-quality results. Moreover, we extend our study to demonstrate the generalizability of the these findings by applying various diffusion samplers, exploring diverse downstream tasks, evaluating post-distilled models, as well as comparing performance relative to training compute. These findings open up new pathways for the development of LDM scaling strategies which can be employed to enhance generative capabilities within limited inference budgets.

Chenyang Qi, Zhengzhong Tu, Keren Ye et al.

Text-driven diffusion models have become increasingly popular for various image editing tasks, including inpainting, stylization, and object replacement. However, it still remains an open research problem to adopt this language-vision paradigm for more fine-level image processing tasks, such as denoising, super-resolution, deblurring, and compression artifact removal. In this paper, we develop SPIRE, a Semantic and restoration Prompt-driven Image Restoration framework that leverages natural language as a user-friendly interface to control the image restoration process. We consider the capacity of prompt information in two dimensions. First, we use content-related prompts to enhance the semantic alignment, effectively alleviating identity ambiguity in the restoration outcomes. Second, our approach is the first framework that supports fine-level instruction through language-based quantitative specification of the restoration strength, without the need for explicit task-specific design. In addition, we introduce a novel fusion mechanism that augments the existing ControlNet architecture by learning to rescale the generative prior, thereby achieving better restoration fidelity. Our extensive experiments demonstrate the superior restoration performance of SPIRE compared to the state of the arts, alongside offering the flexibility of text-based control over the restoration effects.

Kangfu Mei, Hossein Talebi, Mojtaba Ardakani et al.

Single-image super-resolution (SISR) remains challenging due to the inherent difficulty of recovering fine-grained details and preserving perceptual quality from low-resolution inputs. Existing methods often rely on limited image priors, leading to suboptimal results. We propose a novel approach that leverages the rich contextual information available in multiple modalities -- including depth, segmentation, edges, and text prompts -- to learn a powerful generative prior for SISR within a diffusion model framework. We introduce a flexible network architecture that effectively fuses multimodal information, accommodating an arbitrary number of input modalities without requiring significant modifications to the diffusion process. Crucially, we mitigate hallucinations, often introduced by text prompts, by using spatial information from other modalities to guide regional text-based conditioning. Each modality's guidance strength can also be controlled independently, allowing steering outputs toward different directions, such as increasing bokeh through depth or adjusting object prominence via segmentation. Extensive experiments demonstrate that our model surpasses state-of-the-art generative SISR methods, achieving superior visual quality and fidelity. See project page at https://mmsr.kfmei.com/.

Mo Zhou, Keren Ye, Mauricio Delbracio et al.

Real-world image restoration is hampered by diverse degradations stemming from varying capture conditions, capture devices and post-processing pipelines. Existing works make improvements through simulating those degradations and leveraging image generative priors, however generalization to in-the-wild data remains an unresolved problem. In this paper, we focus on complex degradations, i.e., arbitrary mixtures of multiple types of known degradations, which is frequently seen in the wild. A simple yet flexible diffusionbased framework, named UniRes, is proposed to address such degradations in an end-to-end manner. It combines several specialized models during the diffusion sampling steps, hence transferring the knowledge from several well-isolated restoration tasks to the restoration of complex in-the-wild degradations. This only requires well-isolated training data for several degradation types. The framework is flexible as extensions can be added through a unified formulation, and the fidelity-quality trade-off can be adjusted through a new paradigm. Our proposed method is evaluated on both complex-degradation and single-degradation image restoration datasets. Extensive qualitative and quantitative experimental results show consistent performance gain especially for images with complex degradations.

Yuyang Hu, Kangfu Mei, Mojtaba Sahraee-Ardakan et al.

Diffusion models show promise for image restoration, but existing methods often struggle with inconsistent fidelity and undesirable artifacts. To address this, we introduce Kernel Density Steering (KDS), a novel inference-time framework promoting robust, high-fidelity outputs through explicit local mode-seeking. KDS employs an $N$-particle ensemble of diffusion samples, computing patch-wise kernel density estimation gradients from their collective outputs. These gradients steer patches in each particle towards shared, higher-density regions identified within the ensemble. This collective local mode-seeking mechanism, acting as "collective wisdom", steers samples away from spurious modes prone to artifacts, arising from independent sampling or model imperfections, and towards more robust, high-fidelity structures. This allows us to obtain better quality samples at the expense of higher compute by simultaneously sampling multiple particles. As a plug-and-play framework, KDS requires no retraining or external verifiers, seamlessly integrating with various diffusion samplers. Extensive numerical validations demonstrate KDS substantially improves both quantitative and qualitative performance on challenging real-world super-resolution and image inpainting tasks.

Yuyang Hu, Mojtaba Sahraee-Ardakan, Arpit Bansal et al.

Pre-trained image restoration models often fail on real-world, out-of-distribution degradations due to significant domain gaps. Adapting to these unseen domains is challenging, as out-of-distribution data lacks ground truth, and traditional adaptation methods often require complex architectural changes. We propose LEGO (Learning from a Generative Oracle), a practical three-stage framework for post-training domain adaptation without paired data. LEGO converts this unsupervised challenge into a tractable pseudo-supervised one. First, we obtain initial restorations from the pre-trained model. Second, we leverage a frozen, large-scale generative oracle to refine these estimates into high-quality pseudo-ground-truths. Third, we fine-tune the original model using a mixed-supervision strategy combining in-distribution data with these new pseudo-pairs. This approach adapts the model to the new distribution without sacrificing its original robustness or requiring architectural modifications. Experiments demonstrate that LEGO effectively bridges the domain gap, significantly improving performance on diverse real-world benchmarks.

Mojtaba Sahraee-Ardakan, Mauricio Delbracio, Peyman Milanfar

Autonomous (noise-agnostic) generative models, such as Equilibrium Matching and blind diffusion, challenge the standard paradigm by learning a single, time-invariant vector field that operates without explicit noise-level conditioning. While recent work suggests that high-dimensional concentration allows these models to implicitly estimate noise levels from corrupted observations, a fundamental paradox remains: what is the underlying landscape being optimized when the noise level is treated as a random variable, and how can a bounded, noise-agnostic network remain stable near the data manifold where gradients typically diverge? We resolve this paradox by formalizing Marginal Energy, $E_{\text{marg}}(\mathbf{u}) = -\log p(\mathbf{u})$, where $p(\mathbf{u}) = \int p(\mathbf{u}|t)p(t)dt$ is the marginal density of the noisy data integrated over a prior distribution of unknown noise levels. We prove that generation using autonomous models is not merely blind denoising, but a specific form of Riemannian gradient flow on this Marginal Energy. Through a novel relative energy decomposition, we demonstrate that while the raw Marginal Energy landscape possesses a $1/t^p$ singularity normal to the data manifold, the learned time-invariant field implicitly incorporates a local conformal metric that perfectly counteracts the geometric singularity, converting an infinitely deep potential well into a stable attractor. We also establish the structural stability conditions for sampling with autonomous models. We identify a ``Jensen Gap'' in noise-prediction parameterizations that acts as a high-gain amplifier for estimation errors, explaining the catastrophic failure observed in deterministic blind models. Conversely, we prove that velocity-based parameterizations are inherently stable because they satisfy a bounded-gain condition that absorbs posterior uncertainty into a smooth geometric drift.

Keren Ye, Ignacio Garcia Dorado, Michalis Raptis et al.

While recent advancements in Image Super-Resolution (SR) using diffusion models have shown promise in improving overall image quality, their application to scene text images has revealed limitations. These models often struggle with accurate text region localization and fail to effectively model image and multilingual character-to-shape priors. This leads to inconsistencies, the generation of hallucinated textures, and a decrease in the perceived quality of the super-resolved text. To address these issues, we introduce TextSR, a multimodal diffusion model specifically designed for Multilingual Scene Text Image Super-Resolution. TextSR leverages a text detector to pinpoint text regions within an image and then employs Optical Character Recognition (OCR) to extract multilingual text from these areas. The extracted text characters are then transformed into visual shapes using a UTF-8 based text encoder and cross-attention. Recognizing that OCR may sometimes produce inaccurate results in real-world scenarios, we have developed two innovative methods to enhance the robustness of our model. By integrating text character priors with the low-resolution text images, our model effectively guides the super-resolution process, enhancing fine details within the text and improving overall legibility. The superior performance of our model on both the TextZoom and TextVQA datasets sets a new benchmark for STISR, underscoring the efficacy of our approach.

Mo Zhou, Keren Ye, Viraj Shah et al.

Preserving face identity is a critical yet persistent challenge in diffusion-based image restoration. While reference faces offer a path forward, existing reference-based methods often fail to fully exploit their potential. This paper introduces a novel approach that maximizes reference face utility for improved face restoration and identity preservation. Our method makes three key contributions: 1) Composite Context, a comprehensive representation that fuses multi-level (high- and low-level) information from the reference face, offering richer guidance than prior singular representations. 2) Hard Example Identity Loss, a novel loss function that leverages the reference face to address the identity learning inefficiencies found in the existing identity loss. 3) A training-free method to adapt the model to multi-reference inputs during inference. The proposed method demonstrably restores high-quality faces and achieves state-of-the-art identity preserving restoration on benchmarks such as FFHQ-Ref and CelebA-Ref-Test, consistently outperforming previous work.

Dana Weitzner, Mauricio Delbracio, Peyman Milanfar et al.

Recently, diffusion models have gained popularity due to their impressive generative abilities. These models learn the implicit distribution given by the training dataset, and sample new data by transforming random noise through the reverse process, which can be thought of as gradual denoising. In this work, we examine the generation process as a ``correlation machine'', where random noise is repeatedly enhanced in correlation with the implicit given distribution. To this end, we explore the linear case, where the optimal denoiser in the MSE sense is known to be the PCA projection. This enables us to connect the theory of diffusion models to the spiked covariance model, where the dependence of the denoiser on the noise level and the amount of training data can be expressed analytically, in the rank-1 case. In a series of numerical experiments, we extend this result to general low rank data, and show that low frequencies emerge earlier in the generation process, where the denoising basis vectors are more aligned to the true data with a rate depending on their eigenvalues. This model allows us to show that the linear diffusion model converges in mean to the leading eigenvector of the underlying data, similarly to the prevalent power iteration method. Finally, we empirically demonstrate the applicability of our findings beyond the linear case, in the Jacobians of a deep, non-linear denoiser, used in general image generation tasks.

Jay Whang, Mauricio Delbracio, Hossein Talebi et al.

Image deblurring is an ill-posed problem with multiple plausible solutions for a given input image. However, most existing methods produce a deterministic estimate of the clean image and are trained to minimize pixel-level distortion. These metrics are known to be poorly correlated with human perception, and often lead to unrealistic reconstructions. We present an alternative framework for blind deblurring based on conditional diffusion models. Unlike existing techniques, we train a stochastic sampler that refines the output of a deterministic predictor and is capable of producing a diverse set of plausible reconstructions for a given input. This leads to a significant improvement in perceptual quality over existing state-of-the-art methods across multiple standard benchmarks. Our predict-and-refine approach also enables much more efficient sampling compared to typical diffusion models. Combined with a carefully tuned network architecture and inference procedure, our method is competitive in terms of distortion metrics such as PSNR. These results show clear benefits of our diffusion-based method for deblurring and challenge the widely used strategy of producing a single, deterministic reconstruction.

Pascal Tom Getreuer, Peyman Milanfar, Xiyang Luo

Partial differential equations (PDEs) are typically used as models of physical processes but are also of great interest in PDE-based image processing. However, when it comes to their use in imaging, conventional numerical methods for solving PDEs tend to require very fine grid resolution for stability, and as a result have impractically high computational cost. This work applies BLADE (Best Linear Adaptive Enhancement), a shallow learnable filtering framework, to PDE solving, and shows that the resulting approach is efficient and accurate, operating more reliably at coarse grid resolutions than classical methods. As such, the model can be flexibly used for a wide variety of problems in imaging.

Junjie Ke, Qifei Wang, Yilin Wang et al.

Image quality assessment (IQA) is an important research topic for understanding and improving visual experience. The current state-of-the-art IQA methods are based on convolutional neural networks (CNNs). The performance of CNN-based models is often compromised by the fixed shape constraint in batch training. To accommodate this, the input images are usually resized and cropped to a fixed shape, causing image quality degradation. To address this, we design a multi-scale image quality Transformer (MUSIQ) to process native resolution images with varying sizes and aspect ratios. With a multi-scale image representation, our proposed method can capture image quality at different granularities. Furthermore, a novel hash-based 2D spatial embedding and a scale embedding is proposed to support the positional embedding in the multi-scale representation. Experimental results verify that our method can achieve state-of-the-art performance on multiple large scale IQA datasets such as PaQ-2-PiQ, SPAQ and KonIQ-10k.

Innfarn Yoo, Huiwen Chang, Xiyang Luo et al.

Digital watermarking is widely used for copyright protection. Traditional 3D watermarking approaches or commercial software are typically designed to embed messages into 3D meshes, and later retrieve the messages directly from distorted/undistorted watermarked 3D meshes. However, in many cases, users only have access to rendered 2D images instead of 3D meshes. Unfortunately, retrieving messages from 2D renderings of 3D meshes is still challenging and underexplored. We introduce a novel end-to-end learning framework to solve this problem through: 1) an encoder to covertly embed messages in both mesh geometry and textures; 2) a differentiable renderer to render watermarked 3D objects from different camera angles and under varied lighting conditions; 3) a decoder to recover the messages from 2D rendered images. From our experiments, we show that our model can learn to embed information visually imperceptible to humans, and to retrieve the embedded information from 2D renderings that undergo 3D distortions. In addition, we demonstrate that our method can also work with other renderers, such as ray tracers and real-time renderers with and without fine-tuning.

Xiyang Luo, Yinxiao Li, Huiwen Chang et al.

Video watermarking embeds a message into a cover video in an imperceptible manner, which can be retrieved even if the video undergoes certain modifications or distortions. Traditional watermarking methods are often manually designed for particular types of distortions and thus cannot simultaneously handle a broad spectrum of distortions. To this end, we propose a robust deep learning-based solution for video watermarking that is end-to-end trainable. Our model consists of a novel multiscale design where the watermarks are distributed across multiple spatial-temporal scales. It gains robustness against various distortions through a differentiable distortion layer, whereas non-differentiable distortions, such as popular video compression standards, are modeled by a differentiable proxy. Extensive evaluations on a wide variety of distortions show that our method outperforms traditional video watermarking methods as well as deep image watermarking models by a large margin. We further demonstrate the practicality of our method on a realistic video-editing application.

Mangal Prakash, Mauricio Delbracio, Peyman Milanfar et al.

Image denoising and artefact removal are complex inverse problems admitting multiple valid solutions. Unsupervised diversity restoration, that is, obtaining a diverse set of possible restorations given a corrupted image, is important for ambiguity removal in many applications such as microscopy where paired data for supervised training are often unobtainable. In real world applications, imaging noise and artefacts are typically hard to model, leading to unsatisfactory performance of existing unsupervised approaches. This work presents an interpretable approach for unsupervised and diverse image restoration. To this end, we introduce a capable architecture called Hierarchical DivNoising (HDN) based on hierarchical Variational Autoencoder. We show that HDN learns an interpretable multi-scale representation of artefacts and we leverage this interpretability to remove imaging artefacts commonly occurring in microscopy data. Our method achieves state-of-the-art results on twelve benchmark image denoising datasets while providing access to a whole distribution of sensibly restored solutions. Additionally, we demonstrate on three real microscopy datasets that HDN removes artefacts without supervision, being the first method capable of doing so while generating multiple plausible restorations all consistent with the given corrupted image.

Gregory Vaksman, Michael Elad, Peyman Milanfar

The non-local self-similarity property of natural images has been exploited extensively for solving various image processing problems. When it comes to video sequences, harnessing this force is even more beneficial due to the temporal redundancy. In the context of image and video denoising, many classically-oriented algorithms employ self-similarity, splitting the data into overlapping patches, gathering groups of similar ones and processing these together somehow. With the emergence of convolutional neural networks (CNN), the patch-based framework has been abandoned. Most CNN denoisers operate on the whole image, leveraging non-local relations only implicitly by using a large receptive field. This work proposes a novel approach for leveraging self-similarity in the context of video denoising, while still relying on a regular convolutional architecture. We introduce a concept of patch-craft frames - artificial frames that are similar to the real ones, built by tiling matched patches. Our algorithm augments video sequences with patch-craft frames and feeds them to a CNN. We demonstrate the substantial boost in denoising performance obtained with the proposed approach.

Hossein Talebi, Peyman Milanfar

For all the ways convolutional neural nets have revolutionized computer vision in recent years, one important aspect has received surprisingly little attention: the effect of image size on the accuracy of tasks being trained for. Typically, to be efficient, the input images are resized to a relatively small spatial resolution (e.g. 224x224), and both training and inference are carried out at this resolution. The actual mechanism for this re-scaling has been an afterthought: Namely, off-the-shelf image resizers such as bilinear and bicubic are commonly used in most machine learning software frameworks. But do these resizers limit the on task performance of the trained networks? The answer is yes. Indeed, we show that the typical linear resizer can be replaced with learned resizers that can substantially improve performance. Importantly, while the classical resizers typically result in better perceptual quality of the downscaled images, our proposed learned resizers do not necessarily give better visual quality, but instead improve task performance. Our learned image resizer is jointly trained with a baseline vision model. This learned CNN-based resizer creates machine friendly visual manipulations that lead to a consistent improvement of the end task metric over the baseline model. Specifically, here we focus on the classification task with the ImageNet dataset, and experiment with four different models to learn resizers adapted to each model. Moreover, we show that the proposed resizer can also be useful for fine-tuning the classification baselines for other vision tasks. To this end, we experiment with three different baselines to develop image quality assessment (IQA) models on the AVA dataset.

Guy Ohayon, Theo Adrai, Gregory Vaksman et al.

The vast work in Deep Learning (DL) has led to a leap in image denoising research. Most DL solutions for this task have chosen to put their efforts on the denoiser's architecture while maximizing distortion performance. However, distortion driven solutions lead to blurry results with sub-optimal perceptual quality, especially in immoderate noise levels. In this paper we propose a different perspective, aiming to produce sharp and visually pleasing denoised images that are still faithful to their clean sources. Formally, our goal is to achieve high perceptual quality with acceptable distortion. This is attained by a stochastic denoiser that samples from the posterior distribution, trained as a generator in the framework of conditional generative adversarial networks (CGAN). Contrary to distortion-based regularization terms that conflict with perceptual quality, we introduce to the CGAN objective a theoretically founded penalty term that does not force a distortion requirement on individual samples, but rather on their mean. We showcase our proposed method with a novel denoiser architecture that achieves the reformed denoising goal and produces vivid and diverse outcomes in immoderate noise levels.

Juan Carlos Mier, Eddie Huang, Hossein Talebi et al.

Lossy Image compression is necessary for efficient storage and transfer of data. Typically the trade-off between bit-rate and quality determines the optimal compression level. This makes the image quality metric an integral part of any imaging system. While the existing full-reference metrics such as PSNR and SSIM may be less sensitive to perceptual quality, the recently introduced learning methods may fail to generalize to unseen data. In this paper we propose the largest image compression quality dataset to date with human perceptual preferences, enabling the use of deep learning, and we develop a full reference perceptual quality assessment metric for lossy image compression that outperforms the existing state-of-the-art methods. We show that the proposed model can effectively learn from thousands of examples available in the new dataset, and consequently it generalizes better to other unseen datasets of human perceptual preference.

Mauricio Delbracio, Damien Kelly, Michael S. Brown et al.

The first mobile camera phone was sold only 20 years ago, when taking pictures with one's phone was an oddity, and sharing pictures online was unheard of. Today, the smartphone is more camera than phone. How did this happen? This transformation was enabled by advances in computational photography -the science and engineering of making great images from small form factor, mobile cameras. Modern algorithmic and computing advances, including machine learning, have changed the rules of photography, bringing to it new modes of capture, post-processing, storage, and sharing. In this paper, we give a brief history of mobile computational photography and describe some of the key technological components, including burst photography, noise reduction, and super-resolution. At each step, we may draw naive parallels to the human visual system.

Mauricio Delbracio, Ignacio Garcia-Dorado, Sungjoon Choi et al.

We present a highly efficient blind restoration method to remove mild blur in natural images. Contrary to the mainstream, we focus on removing slight blur that is often present, damaging image quality and commonly generated by small out-of-focus, lens blur, or slight camera motion. The proposed algorithm first estimates image blur and then compensates for it by combining multiple applications of the estimated blur in a principled way. To estimate blur we introduce a simple yet robust algorithm based on empirical observations about the distribution of the gradient in sharp natural images. Our experiments show that, in the context of mild blur, the proposed method outperforms traditional and modern blind deblurring methods and runs in a fraction of the time. Our method can be used to blindly correct blur before applying off-the-shelf deep super-resolution methods leading to superior results than other highly complex and computationally demanding techniques. The proposed method estimates and removes mild blur from a 12MP image on a modern mobile phone in a fraction of a second.

Mauricio Delbracio, Hossein Talebi, Peyman Milanfar

Features obtained from object recognition CNNs have been widely used for measuring perceptual similarities between images. Such differentiable metrics can be used as perceptual learning losses to train image enhancement models. However, the choice of the distance function between input and target features may have a consequential impact on the performance of the trained model. While using the norm of the difference between extracted features leads to limited hallucination of details, measuring the distance between distributions of features may generate more textures; yet also more unrealistic details and artifacts. In this paper, we demonstrate that aggregating 1D-Wasserstein distances between CNN activations is more reliable than the existing approaches, and it can significantly improve the perceptual performance of enhancement models. More explicitly, we show that in imaging applications such as denoising, super-resolution, demosaicing, deblurring and JPEG artifact removal, the proposed learning loss outperforms the current state-of-the-art on reference-based perceptual losses. This means that the proposed learning loss can be plugged into different imaging frameworks and produce perceptually realistic results.

Abdullah Abuolaim, Mauricio Delbracio, Damien Kelly et al.

Recent work has shown impressive results on data-driven defocus deblurring using the two-image views available on modern dual-pixel (DP) sensors. One significant challenge in this line of research is access to DP data. Despite many cameras having DP sensors, only a limited number provide access to the low-level DP sensor images. In addition, capturing training data for defocus deblurring involves a time-consuming and tedious setup requiring the camera's aperture to be adjusted. Some cameras with DP sensors (e.g., smartphones) do not have adjustable apertures, further limiting the ability to produce the necessary training data. We address the data capture bottleneck by proposing a procedure to generate realistic DP data synthetically. Our synthesis approach mimics the optical image formation found on DP sensors and can be applied to virtual scenes rendered with standard computer software. Leveraging these realistic synthetic DP images, we introduce a recurrent convolutional network (RCN) architecture that improves deblurring results and is suitable for use with single-frame and multi-frame data (e.g., video) captured by DP sensors. Finally, we show that our synthetic DP data is useful for training DNN models targeting video deblurring applications where access to DP data remains challenging.

Hossein Talebi, Ehsan Amid, Peyman Milanfar et al.

Conducting pairwise comparisons is a widely used approach in curating human perceptual preference data. Typically raters are instructed to make their choices according to a specific set of rules that address certain dimensions of image quality and aesthetics. The outcome of this process is a dataset of sampled image pairs with their associated empirical preference probabilities. Training a model on these pairwise preferences is a common deep learning approach. However, optimizing by gradient descent through mini-batch learning means that the "global" ranking of the images is not explicitly taken into account. In other words, each step of the gradient descent relies only on a limited number of pairwise comparisons. In this work, we demonstrate that regularizing the pairwise empirical probabilities with aggregated rankwise probabilities leads to a more reliable training loss. We show that training a deep image quality assessment model with our rank-smoothed loss consistently improves the accuracy of predicting human preferences.

Qifei Wang, Junjie Ke, Joshua Greaves et al.

Learning multiple domains/tasks with a single model is important for improving data efficiency and lowering inference cost for numerous vision tasks, especially on resource-constrained mobile devices. However, hand-crafting a multi-domain/task model can be both tedious and challenging. This paper proposes a novel approach to automatically learn a multi-path network for multi-domain visual classification on mobile devices. The proposed multi-path network is learned from neural architecture search by applying one reinforcement learning controller for each domain to select the best path in the super-network created from a MobileNetV3-like search space. An adaptive balanced domain prioritization algorithm is proposed to balance optimizing the joint model on multiple domains simultaneously. The determined multi-path model selectively shares parameters across domains in shared nodes while keeping domain-specific parameters within non-shared nodes in individual domain paths. This approach effectively reduces the total number of parameters and FLOPS, encouraging positive knowledge transfer while mitigating negative interference across domains. Extensive evaluations on the Visual Decathlon dataset demonstrate that the proposed multi-path model achieves state-of-the-art performance in terms of accuracy, model size, and FLOPS against other approaches using MobileNetV3-like architectures. Furthermore, the proposed method improves average accuracy over learning single-domain models individually, and reduces the total number of parameters and FLOPS by 78% and 32% respectively, compared to the approach that simply bundles single-domain models for multi-domain learning.

Xiyang Luo, Hossein Talebi, Feng Yang et al.

Handling digital images is almost always accompanied by a lossy compression in order to facilitate efficient transmission and storage. This introduces an unavoidable tension between the allocated bit-budget (rate) and the faithfulness of the resulting image to the original one (distortion). An additional complicating consideration is the effect of the compression on recognition performance by given classifiers (accuracy). This work aims to explore this rate-distortion-accuracy tradeoff. As a case study, we focus on the design of the quantization tables in the JPEG compression standard. We offer a novel optimal tuning of these tables via continuous optimization, leveraging a differential implementation of both the JPEG encoder-decoder and an entropy estimator. This enables us to offer a unified framework that considers the interplay between rate, distortion and classification accuracy. In all these fronts, we report a substantial boost in performance by a simple and easily implemented modification of these tables.

Regev Cohen, Michael Elad, Peyman Milanfar

Inverse problems in image processing are typically cast as optimization tasks, consisting of data-fidelity and stabilizing regularization terms. A recent regularization strategy of great interest utilizes the power of denoising engines. Two such methods are the Plug-and-Play Prior (PnP) and Regularization by Denoising (RED). While both have shown state-of-the-art results in various recovery tasks, their theoretical justification is incomplete. In this paper, we aim to bridge between RED and PnP, enriching the understanding of both frameworks. Towards that end, we reformulate RED as a convex optimization problem utilizing a projection (RED-PRO) onto the fixed-point set of demicontractive denoisers. We offer a simple iterative solution to this problem, by which we show that PnP proximal gradient method is a special case of RED-PRO, while providing guarantees for the convergence of both frameworks to globally optimal solutions. In addition, we present relaxations of RED-PRO that allow for handling denoisers with limited fixed-point sets. Finally, we demonstrate RED-PRO for the tasks of image deblurring and super-resolution, showing improved results with respect to the original RED framework.

Innfarn Yoo, Xiyang Luo, Yilin Wang et al.

Graphics Interchange Format (GIF) is a widely used image file format. Due to the limited number of palette colors, GIF encoding often introduces color banding artifacts. Traditionally, dithering is applied to reduce color banding, but introducing dotted-pattern artifacts. To reduce artifacts and provide a better and more efficient GIF encoding, we introduce a differentiable GIF encoding pipeline, which includes three novel neural networks: PaletteNet, DitherNet, and BandingNet. Each of these three networks provides an important functionality within the GIF encoding pipeline. PaletteNet predicts a near-optimal color palette given an input image. DitherNet manipulates the input image to reduce color banding artifacts and provides an alternative to traditional dithering. Finally, BandingNet is designed to detect color banding, and provides a new perceptual loss specifically for GIF images. As far as we know, this is the first fully differentiable GIF encoding pipeline based on deep neural networks and compatible with existing GIF decoders. User study shows that our algorithm is better than Floyd-Steinberg based GIF encoding.

David Berthelot, Peyman Milanfar, Ian Goodfellow

Generating realistic images is difficult, and many formulations for this task have been proposed recently. If we restrict the task to that of generating a particular class of images, however, the task becomes more tractable. That is to say, instead of generating an arbitrary image as a sample from the manifold of natural images, we propose to sample images from a particular "subspace" of natural images, directed by a low-resolution image from the same subspace. The problem we address, while close to the formulation of the single-image super-resolution problem, is in fact rather different. Single image super-resolution is the task of predicting the image closest to the ground truth from a relatively low resolution image. We propose to produce samples of high resolution images given extremely small inputs with a new method called Latent Adversarial Generator (LAG). In our generative sampling framework, we only use the input (possibly of very low-resolution) to direct what class of samples the network should produce. As such, the output of our algorithm is not a unique image that relates to the input, but rather a possible se} of related images sampled from the manifold of natural images. Our method learns exclusively in the latent space of the adversary using perceptual loss -- it does not have a pixel loss.

Xiang Zhu, Hossein Talebi, Xinwei Shi et al.

In machine learning based single image super-resolution, the degradation model is embedded in training data generation. However, most existing satellite image super-resolution methods use a simple down-sampling model with a fixed kernel to create training images. These methods work fine on synthetic data, but do not perform well on real satellite images. We propose a realistic training data generation model for commercial satellite imagery products, which includes not only the imaging process on satellites but also the post-process on the ground. We also propose a convolutional neural network optimized for satellite images. Experiments show that the proposed training data generation model is able to improve super-resolution performance on real satellite images.

Ignacio Garcia-Dorado, Pascal Getreuer, Bartlomiej Wronski et al.

We present a framework for interactive design of new image stylizations using a wide range of predefined filter blocks. Both novel and off-the-shelf image filtering and rendering techniques are extended and combined to allow the user to unleash their creativity to intuitively invent, modify, and tune new styles from a given set of filters. In parallel to this manual design, we propose a novel procedural approach that automatically assembles sequences of filters, leading to unique and novel styles. An important aim of our framework is to allow for interactive exploration and design, as well as to enable videos and camera streams to be stylized on the fly. In order to achieve this real-time performance, we use the \textit{Best Linear Adaptive Enhancement} (BLADE) framework -- an interpretable shallow machine learning method that simulates complex filter blocks in real time. Our representative results include over a dozen styles designed using our interactive tool, a set of styles created procedurally, and new filters trained with our BLADE approach.

Hossein Talebi, Damien Kelly, Xiyang Luo et al.

Could we compress images via standard codecs while avoiding visible artifacts? The answer is obvious -- this is doable as long as the bit budget is generous enough. What if the allocated bit-rate for compression is insufficient? Then unfortunately, artifacts are a fact of life. Many attempts were made over the years to fight this phenomenon, with various degrees of success. In this work we aim to break the unholy connection between bit-rate and image quality, and propose a way to circumvent compression artifacts by pre-editing the incoming image and modifying its content to fit the given bits. We design this editing operation as a learned convolutional neural network, and formulate an optimization problem for its training. Our loss takes into account a proximity between the original image and the edited one, a bit-budget penalty over the proposed image, and a no-reference image quality measure for forcing the outcome to be visually pleasing. The proposed approach is demonstrated on the popular JPEG compression, showing savings in bits and/or improvements in visual quality, obtained with intricate editing effects.

Xiyang Luo, Ruohan Zhan, Huiwen Chang et al.

Watermarking is the process of embedding information into an image that can survive under distortions, while requiring the encoded image to have little or no perceptual difference from the original image. Recently, deep learning-based methods achieved impressive results in both visual quality and message payload under a wide variety of image distortions. However, these methods all require differentiable models for the image distortions at training time, and may generalize poorly to unknown distortions. This is undesirable since the types of distortions applied to watermarked images are usually unknown and non-differentiable. In this paper, we propose a new framework for distortion-agnostic watermarking, where the image distortion is not explicitly modeled during training. Instead, the robustness of our system comes from two sources: adversarial training and channel coding. Compared to training on a fixed set of distortions and noise levels, our method achieves comparable or better results on distortions available during training, and better performance on unknown distortions.

Gregory Vaksman, Michael Elad, Peyman Milanfar

Image denoising is a well studied problem with an extensive activity that has spread over several decades. Despite the many available denoising algorithms, the quest for simple, powerful and fast denoisers is still an active and vibrant topic of research. Leading classical denoising methods are typically designed to exploit the inner structure in images by modeling local overlapping patches, while operating in an unsupervised fashion. In contrast, recent newcomers to this arena are supervised and universal neural-network-based methods that bypass this modeling altogether, targeting the inference goal directly and globally, while tending to be very deep and parameter heavy. This work proposes a novel lightweight learnable architecture for image denoising, and presents a combination of supervised and unsupervised training of it, the first aiming for a universal denoiser and the second for adapting it to the incoming image. Our architecture embeds in it several of the main concepts taken from classical methods, relying on patch processing, leveraging non-local self-similarity, exploiting representation sparsity and providing a multiscale treatment. Our proposed universal denoiser achieves near state-of-the-art results, while using a small fraction of the typical number of parameters. In addition, we introduce and demonstrate two highly effective ways for further boosting the denoising performance, by adapting this universal network to the input image.