Convolutional neural networks based single-image super-resolution (SISR) has made great progress in recent years. However, it is difficult to apply these methods to real-world scenarios due to the computational and memory cost. Meanwhile, how to take full advantage of the intermediate features under the constraints of limited parameters and calculations is also a huge challenge. To alleviate these issues, we propose a lightweight yet efficient Feature Distillation Interaction Weighted Network (FDIWN). Specifically, FDIWN utilizes a series of specially designed Feature Shuffle Weighted Groups (FSWG) as the backbone, and several novel mutual Wide-residual Distillation Interaction Blocks (WDIB) form an FSWG. In addition, Wide Identical Residual Weighting (WIRW) units and Wide Convolutional Residual Weighting (WCRW) units are introduced into WDIB for better feature distillation. Moreover, a Wide-Residual Distillation Connection (WRDC) framework and a Self-Calibration Fusion (SCF) unit are proposed to interact features with different scales more flexibly and efficiently.Extensive experiments show that our FDIWN is superior to other models to strike a good balance between model performance and efficiency. The code is available at https://github.com/IVIPLab/FDIWN.
Burst super-resolution (SR) technique provides a possibility of restoring rich details from low-quality images. However, since real world low-resolution (LR) images in practical applications have multiple complicated and unknown degradations, existing non-blind (e.g., bicubic) designed networks usually suffer severe performance drop in recovering high-resolution (HR) images. In this paper, we address the problem of reconstructing HR images from raw burst sequences acquired from a modern handheld device. The central idea is a kernel-guided strategy which can solve the burst SR problem with two steps: kernel estimation and HR image restoration. The former estimates burst kernels from raw inputs, while the latter predicts the super-resolved image based on the estimated kernels. Furthermore, we introduce a pyramid kernel-aware deformable alignment module which can effectively align the raw images with consideration of the blurry priors. Extensive experiments on synthetic and real-world datasets demonstrate that the proposed method can perform favorable state-of-the-art performance in the burst SR problem. Our codes are available at \url{https://github.com/shermanlian/KBNet}.
Event cameras are novel bio-inspired sensors that measure per-pixel brightness differences asynchronously. Recovering brightness from events is appealing since the reconstructed images inherit the high dynamic range (HDR) and high-speed properties of events; hence they can be used in many robotic vision applications and to generate slow-motion HDR videos. However, state-of-the-art methods tackle this problem by training an event-to-image Recurrent Neural Network (RNN), which lacks explainability and is difficult to tune. In this work we show, for the first time, how tackling the combined problem of motion and brightness estimation leads us to formulate event-based image reconstruction as a linear inverse problem that can be solved without training an image reconstruction RNN. Instead, classical and learning-based regularizers are used to solve the problem and remove artifacts from the reconstructed images. The experiments show that the proposed approach generates images with visual quality on par with state-of-the-art methods despite only using data from a short time interval. State-of-the-art results are achieved using an image denoising Convolutional Neural Network (CNN) as the regularization function. The proposed regularized formulation and solvers have a unifying character because they can be applied also to reconstruct brightness from the second derivative. Additionally, the formulation is attractive because it can be naturally combined with super-resolution, motion-segmentation and color demosaicing. Code is available at https://github.com/tub-rip/event_based_image_rec_inverse_problem
Nowadays, there is an explosive growth of screen contents due to the wide application of screen sharing, remote cooperation, and online education. To match the limited terminal bandwidth, high-resolution (HR) screen contents may be downsampled and compressed. At the receiver side, the super-resolution (SR) of low-resolution (LR) screen content images (SCIs) is highly demanded by the HR display or by the users to zoom in for detail observation. However, image SR methods mostly designed for natural images do not generalize well for SCIs due to the very different image characteristics as well as the requirement of SCI browsing at arbitrary scales. To this end, we propose a novel Implicit Transformer Super-Resolution Network (ITSRN) for SCISR. For high-quality continuous SR at arbitrary ratios, pixel values at query coordinates are inferred from image features at key coordinates by the proposed implicit transformer and an implicit position encoding scheme is proposed to aggregate similar neighboring pixel values to the query one. We construct benchmark SCI1K and SCI1K-compression datasets with LR and HR SCI pairs. Extensive experiments show that the proposed ITSRN significantly outperforms several competitive continuous and discrete SR methods for both compressed and uncompressed SCIs.
Deep learning methods have shown outstanding performance in many applications, including single-image super-resolution (SISR). With residual connection architecture, deeply stacked convolutional neural networks provide a substantial performance boost for SISR, but their huge parameters and computational loads are impractical for real-world applications. Thus, designing lightweight models with acceptable performance is one of the major tasks in current SISR research. The objective of lightweight network design is to balance a computational load and reconstruction performance. Most of the previous methods have manually designed complex and predefined fixed structures, which generally required a large number of experiments and lacked flexibility in the diversity of input image statistics. In this paper, we propose a dynamic residual self-attention network (DRSAN) for lightweight SISR, while focusing on the automated design of residual connections between building blocks. The proposed DRSAN has dynamic residual connections based on dynamic residual attention (DRA), which adaptively changes its structure according to input statistics. Specifically, we propose a dynamic residual module that explicitly models the DRA by finding the interrelation between residual paths and input image statistics, as well as assigning proper weights to each residual path. We also propose a residual self-attention (RSA) module to further boost the performance, which produces 3-dimensional attention maps without additional parameters by cooperating with residual structures. The proposed dynamic scheme, exploiting the combination of DRA and RSA, shows an efficient trade-off between computational complexity and network performance. Experimental results show that the DRSAN performs better than or comparable to existing state-of-the-art lightweight models for SISR.
Convolutional neural networks have turned into an illustrious tool for image fusion and super-resolution. However, their excellent performance cannot work without large fixed-paired datasets; and additionally, these high-demanded ground truth data always cannot be obtained easily in fusion tasks. In this study, we show that, the structures of generative networks capture a great deal of image feature priors, and then these priors are sufficient to reconstruct high-quality fused super-resolution result using only low-resolution inputs. By this way, we propose a novel self-supervised dataset-free method for adaptive infrared (IR) and visible (VIS) image super-resolution fusion named Deep Retinex Fusion (DRF). The key idea of DRF is first generating component priors which are disentangled from physical model using our designed generative networks ZipperNet, LightingNet and AdjustingNet, then combining these priors which captured by networks via adaptive fusion loss functions based on Retinex theory, and finally reconstructing the super-resolution fusion results. Furthermore, in order to verify the effectiveness of our reported DRF, both qualitative and quantitative experiments via comparing with other state-of-the-art methods are performed using different test sets. These results prove that, comparing with large datasets trained methods, DRF which works without any dataset achieves the best super-resolution fusion performance; and more importantly, DRF can adaptively balance IR and VIS information and has good noise immunity. DRF codes are open source available at https://github.com/GuYuanjie/Deep-Retinex-fusion.
Different from the Single Image Super-Resolution(SISR) task, the key for Video Super-Resolution(VSR) task is to make full use of complementary information across frames to reconstruct the high-resolution sequence. Since images from different frames with diverse motion and scene, accurately aligning multiple frames and effectively fusing different frames has always been the key research work of VSR tasks. To utilize rich complementary information of neighboring frames, in this paper, we propose a multi-stage VSR deep architecture, dubbed as PP-MSVSR, with local fusion module, auxiliary loss and re-align module to refine the enhanced result progressively. Specifically, in order to strengthen the fusion of features across frames in feature propagation, a local fusion module is designed in stage-1 to perform local feature fusion before feature propagation. Moreover, we introduce an auxiliary loss in stage-2 to make the features obtained by the propagation module reserve more correlated information connected to the HR space, and introduce a re-align module in stage-3 to make full use of the feature information of the previous stage. Extensive experiments substantiate that PP-MSVSR achieves a promising performance of Vid4 datasets, which achieves a PSNR of 28.13dB with only 1.45M parameters. And the PP-MSVSR-L exceeds all state of the art method on REDS4 datasets with considerable parameters. Code and models will be released in PaddleGAN\footnote{https://github.com/PaddlePaddle/PaddleGAN.}.
Denoising diffusion probabilistic models have recently received much research attention since they outperform alternative approaches, such as GANs, and currently provide state-of-the-art generative performance. The superior performance of diffusion models has made them an appealing tool in several applications, including inpainting, super-resolution, and semantic editing. In this paper, we demonstrate that diffusion models can also serve as an instrument for semantic segmentation, especially in the setup when labeled data is scarce. In particular, for several pretrained diffusion models, we investigate the intermediate activations from the networks that perform the Markov step of the reverse diffusion process. We show that these activations effectively capture the semantic information from an input image and appear to be excellent pixel-level representations for the segmentation problem. Based on these observations, we describe a simple segmentation method, which can work even if only a few training images are provided. Our approach significantly outperforms the existing alternatives on several datasets for the same amount of human supervision.
Despite the potential of neural scene representations to effectively compress 3D scalar fields at high reconstruction quality, the computational complexity of the training and data reconstruction step using scene representation networks limits their use in practical applications. In this paper, we analyze whether scene representation networks can be modified to reduce these limitations and whether such architectures can also be used for temporal reconstruction tasks. We propose a novel design of scene representation networks using GPU tensor cores to integrate the reconstruction seamlessly into on-chip raytracing kernels, and compare the quality and performance of this network to alternative network- and non-network-based compression schemes. The results indicate competitive quality of our design at high compression rates, and significantly faster decoding times and lower memory consumption during data reconstruction. We investigate how density gradients can be computed using the network and show an extension where density, gradient and curvature are predicted jointly. As an alternative to spatial super-resolution approaches for time-varying fields, we propose a solution that builds upon latent-space interpolation to enable random access reconstruction at arbitrary granularity. We summarize our findings in the form of an assessment of the strengths and limitations of scene representation networks \changed{for compression domain volume rendering, and outline future research directions.
The onset of rheumatic diseases such as rheumatoid arthritis is typically subclinical, which results in challenging early detection of the disease. However, characteristic changes in the anatomy can be detected using imaging techniques such as MRI or CT. Modern imaging techniques such as chemical exchange saturation transfer (CEST) MRI drive the hope to improve early detection even further through the imaging of metabolites in the body. To image small structures in the joints of patients, typically one of the first regions where changes due to the disease occur, a high resolution for the CEST MR imaging is necessary. Currently, however, CEST MR suffers from an inherently low resolution due to the underlying physical constraints of the acquisition. In this work we compared established up-sampling techniques to neural network-based super-resolution approaches. We could show, that neural networks are able to learn the mapping from low-resolution to high-resolution unsaturated CEST images considerably better than present methods. On the test set a PSNR of 32.29dB (+10%), a NRMSE of 0.14 (+28%), and a SSIM of 0.85 (+15%) could be achieved using a ResNet neural network, improving the baseline considerably. This work paves the way for the prospective investigation of neural networks for super-resolution CEST MRI and, followingly, might lead to a earlier detection of the onset of rheumatic diseases.
Recent works show that convolutional neural network (CNN) architectures have a spectral bias towards lower frequencies, which has been leveraged for various image restoration tasks in the Deep Image Prior (DIP) framework. The benefit of the inductive bias the network imposes in the DIP framework depends on the architecture. Therefore, researchers have studied how to automate the search to determine the best-performing model. However, common neural architecture search (NAS) techniques are resource and time-intensive. Moreover, best-performing models are determined for a whole dataset of images instead of for each image independently, which would be prohibitively expensive. In this work, we first show that optimal neural architectures in the DIP framework are image-dependent. Leveraging this insight, we then propose an image-specific NAS strategy for the DIP framework that requires substantially less training than typical NAS approaches, effectively enabling image-specific NAS. We justify the proposed strategy's effectiveness by (1) demonstrating its performance on a NAS Dataset for DIP that includes 522 models from a particular search space (2) conducting extensive experiments on image denoising, inpainting, and super-resolution tasks. Our experiments show that image-specific metrics can reduce the search space to a small cohort of models, of which the best model outperforms current NAS approaches for image restoration. Codes and datasets are available at https://github.com/ozgurkara99/ISNAS-DIP.
Long-range temporal alignment is critical yet challenging for video restoration tasks. Recently, some works attempt to divide the long-range alignment into several sub-alignments and handle them progressively. Although this operation is helpful in modeling distant correspondences, error accumulation is inevitable due to the propagation mechanism. In this work, we present a novel, generic iterative alignment module which employs a gradual refinement scheme for sub-alignments, yielding more accurate motion compensation. To further enhance the alignment accuracy and temporal consistency, we develop a non-parametric re-weighting method, where the importance of each neighboring frame is adaptively evaluated in a spatial-wise way for aggregation. By virtue of the proposed strategies, our model achieves state-of-the-art performance on multiple benchmarks across a range of video restoration tasks including video super-resolution, denoising and deblurring. Our project is available in \url{https://github.com/redrock303/Revisiting-Temporal-Alignment-for-Video-Restoration.git}.
With the development of Deep Neural Networks (DNNs), plenty of methods based on DNNs have been proposed for Single Image Super-Resolution (SISR). However, existing methods mostly train the DNNs on uniformly sampled LR-HR patch pairs, which makes them fail to fully exploit informative patches within the image. In this paper, we present a simple yet effective data augmentation method. We first devise a heuristic metric to evaluate the informative importance of each patch pair. In order to reduce the computational cost for all patch pairs, we further propose to optimize the calculation of our metric by integral image, achieving about two orders of magnitude speedup. The training patch pairs are sampled according to their informative importance with our method. Extensive experiments show our sampling augmentation can consistently improve the convergence and boost the performance of various SISR architectures, including EDSR, RCAN, RDN, SRCNN and ESPCN across different scaling factors (x2, x3, x4). Code is available at https://github.com/littlepure2333/SamplingAug
Contrastive learning has achieved remarkable success on various high-level tasks, but there are fewer contrastive learning-based methods proposed for low-level tasks. It is challenging to adopt vanilla contrastive learning technologies proposed for high-level visual tasks to low-level image restoration problems straightly. Because the acquired high-level global visual representations are insufficient for low-level tasks requiring rich texture and context information. In this paper, we investigate the contrastive learning-based single image super-resolution from two perspectives: positive and negative sample construction and feature embedding. The existing methods take naive sample construction approaches (e.g., considering the low-quality input as a negative sample and the ground truth as a positive sample) and adopt a prior model (e.g., pre-trained VGG model) to obtain the feature embedding. To this end, we propose a practical contrastive learning framework for SISR, named PCL-SR. We involve the generation of many informative positive and hard negative samples in frequency space. Instead of utilizing an additional pre-trained network, we design a simple but effective embedding network inherited from the discriminator network which is more task-friendly. Compared with existing benchmark methods, we re-train them by our proposed PCL-SR framework and achieve superior performance. Extensive experiments have been conducted to show the effectiveness and technical contributions of our proposed PCL-SR thorough ablation studies. The code and pre-trained models can be found at https://github.com/Aitical/PCL-SISR.
Normalization like Batch Normalization (BN) is a milestone technique to normalize the distributions of intermediate layers in deep learning, enabling faster training and better generalization accuracy. However, in fidelity image Super-Resolution (SR), it is believed that normalization layers get rid of range flexibility by normalizing the features and they are simply removed from modern SR networks. In this paper, we study this phenomenon quantitatively and qualitatively. We found that the standard deviation of the residual feature shrinks a lot after normalization layers, which causes the performance degradation in SR networks. Standard deviation reflects the amount of variation of pixel values. When the variation becomes smaller, the edges will become less discriminative for the network to resolve. To address this problem, we propose an Adaptive Deviation Modulator (AdaDM), in which a modulation factor is adaptively predicted to amplify the pixel deviation. For better generalization performance, we apply BN in state-of-the-art SR networks with the proposed AdaDM. Meanwhile, the deviation amplification strategy in AdaDM makes the edge information in the feature more distinguishable. As a consequence, SR networks with BN and our AdaDM can get substantial performance improvements on benchmark datasets. Extensive experiments have been conducted to show the effectiveness of our method.
The diversity and complexity of degradations in real-world video super-resolution (VSR) pose non-trivial challenges in inference and training. First, while long-term propagation leads to improved performance in cases of mild degradations, severe in-the-wild degradations could be exaggerated through propagation, impairing output quality. To balance the tradeoff between detail synthesis and artifact suppression, we found an image pre-cleaning stage indispensable to reduce noises and artifacts prior to propagation. Equipped with a carefully designed cleaning module, our RealBasicVSR outperforms existing methods in both quality and efficiency. Second, real-world VSR models are often trained with diverse degradations to improve generalizability, requiring increased batch size to produce a stable gradient. Inevitably, the increased computational burden results in various problems, including 1) speed-performance tradeoff and 2) batch-length tradeoff. To alleviate the first tradeoff, we propose a stochastic degradation scheme that reduces up to 40\% of training time without sacrificing performance. We then analyze different training settings and suggest that employing longer sequences rather than larger batches during training allows more effective uses of temporal information, leading to more stable performance during inference. To facilitate fair comparisons, we propose the new VideoLQ dataset, which contains a large variety of real-world low-quality video sequences containing rich textures and patterns. Our dataset can serve as a common ground for benchmarking. Code, models, and the dataset will be made publicly available.
We study joint video and language (VL) pre-training to enable cross-modality learning and benefit plentiful downstream VL tasks. Existing works either extract low-quality video features or learn limited text embedding, while neglecting that high-resolution videos and diversified semantics can significantly improve cross-modality learning. In this paper, we propose a novel High-resolution and Diversified VIdeo-LAnguage pre-training model (HD-VILA) for many visual tasks. In particular, we collect a large dataset with two distinct properties: 1) the first high-resolution dataset including 371.5k hours of 720p videos, and 2) the most diversified dataset covering 15 popular YouTube categories. To enable VL pre-training, we jointly optimize the HD-VILA model by a hybrid Transformer that learns rich spatiotemporal features, and a multimodal Transformer that enforces interactions of the learned video features with diversified texts. Our pre-training model achieves new state-of-the-art results in 10 VL understanding tasks and 2 more novel text-to-visual generation tasks. For example, we outperform SOTA models with relative increases of 40.4% R@1 in zero-shot MSR-VTT text-to-video retrieval task and 55.4% in high-resolution dataset LSMDC. The learned VL embedding is also effective in generating visually pleasing and semantically relevant results in text-to-visual editing and super-resolution tasks.
Recent works with an implicit neural function shed light on representing images in arbitrary resolution. However, a standalone multi-layer perceptron shows limited performance in learning high-frequency components. In this paper, we propose a Local Texture Estimator (LTE), a dominant-frequency estimator for natural images, enabling an implicit function to capture fine details while reconstructing images in a continuous manner. When jointly trained with a deep super-resolution (SR) architecture, LTE is capable of characterizing image textures in 2D Fourier space. We show that an LTE-based neural function achieves favorable performance against existing deep SR methods within an arbitrary-scale factor. Furthermore, we demonstrate that our implementation takes the shortest running time compared to previous works.
Recently, deep-learning-based super-resolution methods have achieved excellent performances, but mainly focus on training a single generalized deep network by feeding numerous samples. Yet intuitively, each image has its representation, and is expected to acquire an adaptive model. For this issue, we propose a novel image-specific convolutional kernel modulation (IKM) by exploiting the global contextual information of image or feature to generate an attention weight for adaptively modulating the convolutional kernels, which outperforms the vanilla convolution and several existing attention mechanisms while embedding into the state-of-the-art architectures without any additional parameters. Particularly, to optimize our IKM in mini-batch training, we introduce an image-specific optimization (IsO) algorithm, which is more effective than the conventional mini-batch SGD optimization. Furthermore, we investigate the effect of IKM on the state-of-the-art architectures and exploit a new backbone with U-style residual learning and hourglass dense block learning, terms U-Hourglass Dense Network (U-HDN), which is an appropriate architecture to utmost improve the effectiveness of IKM theoretically and experimentally. Extensive experiments on single image super-resolution show that the proposed methods achieve superior performances over state-of-the-art methods. Code is available at github.com/YuanfeiHuang/IKM.
In recent years, there has been a growing interest in deep learning-based pansharpening. Thus far, research has mainly focused on architectures. Nonetheless, model training is an equally important issue. A first problem is the absence of ground truths, unavoidable in pansharpening. This is often addressed by training networks in a reduced resolution domain and using the original data as ground truth, relying on an implicit scale invariance assumption. However, on full resolution images results are often disappointing, suggesting such invariance not to hold. A further problem is the scarcity of training data, which causes a limited generalization ability and a poor performance on off-training test images. In this paper, we propose a full-resolution training framework for deep learning-based pansharpening. The framework is fully general and can be used for any deep learning-based pansharpening model. Training takes place in the high-resolution domain, relying only on the original data, thus avoiding any loss of information. To ensure spectral and spatial fidelity, a suitable two-component loss is defined. The spectral component enforces consistency between the pansharpened output and the low-resolution multispectral input. The spatial component, computed at high-resolution, maximizes the local correlation between each pansharpened band and the panchromatic input. At testing time, the target-adaptive operating modality is adopted, achieving good generalization with a limited computational overhead. Experiments carried out on WorldView-3, WorldView-2, and GeoEye-1 images show that methods trained with the proposed framework guarantee a pretty good performance in terms of both full-resolution numerical indexes and visual quality.