PyTorchKR Papers

최신 논문

총 1,320편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2201.12288공식 코드 · Papers with Code

VRT: A Video Restoration Transformer

Video restoration (e.g., video super-resolution) aims to restore high-quality frames from low-quality frames. Different from single image restoration, video restoration generally requires to utilize temporal information from multiple adjacent but usually misaligned video frames. Existing deep methods generally tackle with this by exploiting a sliding window strategy or a recurrent architecture, which either is restricted by frame-by-frame restoration or lacks long-range modelling ability. In this paper, we propose a Video Restoration Transformer (VRT) with parallel frame prediction and long-range temporal dependency modelling abilities. More specifically, VRT is composed of multiple scales, each of which consists of two kinds of modules: temporal mutual self attention (TMSA) and parallel warping. TMSA divides the video into small clips, on which mutual attention is applied for joint motion estimation, feature alignment and feature fusion, while self attention is used for feature extraction. To enable cross-clip interactions, the video sequence is shifted for every other layer. Besides, parallel warping is used to further fuse information from neighboring frames by parallel feature warping. Experimental results on five tasks, including video super-resolution, video deblurring, video denoising, video frame interpolation and space-time video super-resolution, demonstrate that VRT outperforms the state-of-the-art methods by large margins ($\textbf{up to 2.16dB}$) on fourteen benchmark datasets.

arXiv:2201.11279공식 코드 · Papers with Code

Revisiting RCAN: Improved Training for Image Super-Resolution

Image super-resolution (SR) is a fast-moving field with novel architectures attracting the spotlight. However, most SR models were optimized with dated training strategies. In this work, we revisit the popular RCAN model and examine the effect of different training options in SR. Surprisingly (or perhaps as expected), we show that RCAN can outperform or match nearly all the CNN-based SR architectures published after RCAN on standard benchmarks with a proper training strategy and minimal architecture change. Besides, although RCAN is a very large SR architecture with more than four hundred convolutional layers, we draw a notable conclusion that underfitting is still the main problem restricting the model capability instead of overfitting. We observe supportive evidence that increasing training iterations clearly improves the model performance while applying regularization techniques generally degrades the predictions. We denote our simply revised RCAN as RCAN-it and recommend practitioners to use it as baselines for future research. Code is publicly available at https://github.com/zudi-lin/rcan-it.

arXiv:2201.10747공식 코드 · Papers with Code

Learning Multiple Probabilistic Degradation Generators for Unsupervised Real World Image Super Resolution

Unsupervised real world super resolution (USR) aims to restore high-resolution (HR) images given low-resolution (LR) inputs, and its difficulty stems from the absence of paired dataset. One of the most common approaches is synthesizing noisy LR images using GANs (i.e., degradation generators) and utilizing a synthetic dataset to train the model in a supervised manner. Although the goal of training the degradation generator is to approximate the distribution of LR images given a HR image, previous works have heavily relied on the unrealistic assumption that the conditional distribution is a delta function and learned the deterministic mapping from the HR image to a LR image. In this paper, we show that we can improve the performance of USR models by relaxing the assumption and propose to train the probabilistic degradation generator. Our probabilistic degradation generator can be viewed as a deep hierarchical latent variable model and is more suitable for modeling the complex conditional distribution. We also reveal the notable connection with the noise injection of StyleGAN. Furthermore, we train multiple degradation generators to improve the mode coverage and apply collaborative learning for ease of training. We outperform several baselines on benchmark datasets in terms of PSNR and SSIM and demonstrate the robustness of our method on unseen data distribution. Code is available at https://github.com/sangyun884/MSSR.

arXiv:2201.09851공식 코드 · Papers with Code

Hyperspectral Image Super-resolution with Deep Priors and Degradation Model Inversion

To overcome inherent hardware limitations of hyperspectral imaging systems with respect to their spatial resolution, fusion-based hyperspectral image (HSI) super-resolution is attracting increasing attention. This technique aims to fuse a low-resolution (LR) HSI and a conventional high-resolution (HR) RGB image in order to obtain an HR HSI. Recently, deep learning architectures have been used to address the HSI super-resolution problem and have achieved remarkable performance. However, they ignore the degradation model even though this model has a clear physical interpretation and may contribute to improve the performance. We address this problem by proposing a method that, on the one hand, makes use of the linear degradation model in the data-fidelity term of the objective function and, on the other hand, utilizes the output of a convolutional neural network for designing a deep prior regularizer in spectral and spatial gradient domains. Experiments show the performance improvement achieved with this strategy.

arXiv:2201.09314공식 코드 · Papers with Code

Perceptual cGAN for MRI Super-resolution

Capturing high-resolution magnetic resonance (MR) images is a time consuming process, which makes it unsuitable for medical emergencies and pediatric patients. Low-resolution MR imaging, by contrast, is faster than its high-resolution counterpart, but it compromises on fine details necessary for a more precise diagnosis. Super-resolution (SR), when applied to low-resolution MR images, can help increase their utility by synthetically generating high-resolution images with little additional time. In this paper, we present a SR technique for MR images that is based on generative adversarial networks (GANs), which have proven to be quite useful in generating sharp-looking details in SR. We introduce a conditional GAN with perceptual loss, which is conditioned upon the input low-resolution image, which improves the performance for isotropic and anisotropic MRI super-resolution.

arXiv:2201.07422공식 코드 · Papers with Code

Self-Supervised Deep Blind Video Super-Resolution

Existing deep learning-based video super-resolution (SR) methods usually depend on the supervised learning approach, where the training data is usually generated by the blurring operation with known or predefined kernels (e.g., Bicubic kernel) followed by a decimation operation. However, this does not hold for real applications as the degradation process is complex and cannot be approximated by these idea cases well. Moreover, obtaining high-resolution (HR) videos and the corresponding low-resolution (LR) ones in real-world scenarios is difficult. To overcome these problems, we propose a self-supervised learning method to solve the blind video SR problem, which simultaneously estimates blur kernels and HR videos from the LR videos. As directly using LR videos as supervision usually leads to trivial solutions, we develop a simple and effective method to generate auxiliary paired data from original LR videos according to the image formation of video SR, so that the networks can be better constrained by the generated paired data for both blur kernel estimation and latent HR video restoration. In addition, we introduce an optical flow estimation module to exploit the information from adjacent frames for HR video restoration. Experiments show that our method performs favorably against state-of-the-art ones on benchmarks and real-world videos.

arXiv:2201.06045공식 코드 · Papers with Code

CISRNet: Compressed Image Super-Resolution Network

In recent years, tons of research has been conducted on Single Image Super-Resolution (SISR). However, to the best of our knowledge, few of these studies are mainly focused on compressed images. A problem such as complicated compression artifacts hinders the advance of this study in spite of its high practical values. To tackle this problem, we proposed CISRNet; a network that employs a two-stage coarse-to-fine learning framework that is mainly optimized for Compressed Image Super-Resolution Problem. Specifically, CISRNet consists of two main subnetworks; the coarse and refinement network, where recursive and residual learning is employed within these two networks respectively. Extensive experiments show that with a careful design choice, CISRNet performs favorably against competing Single-Image Super-Resolution methods in the Compressed Image Super-Resolution tasks.

arXiv:2201.04898공식 코드 · Papers with Code

Flexible Style Image Super-Resolution using Conditional Objective

Recent studies have significantly enhanced the performance of single-image super-resolution (SR) using convolutional neural networks (CNNs). While there can be many high-resolution (HR) solutions for a given input, most existing CNN-based methods do not explore alternative solutions during the inference. A typical approach to obtaining alternative SR results is to train multiple SR models with different loss weightings and exploit the combination of these models. Instead of using multiple models, we present a more efficient method to train a single adjustable SR model on various combinations of losses by taking advantage of multi-task learning. Specifically, we optimize an SR model with a conditional objective during training, where the objective is a weighted sum of multiple perceptual losses at different feature levels. The weights vary according to given conditions, and the set of weights is defined as a style controller. Also, we present an architecture appropriate for this training scheme, which is the Residual-in-Residual Dense Block equipped with spatial feature transformation layers. At the inference phase, our trained model can generate locally different outputs conditioned on the style control map. Extensive experiments show that the proposed SR model produces various desirable reconstructions without artifacts and yields comparable quantitative performance to state-of-the-art SR methods.

arXiv:2201.04402공식 코드 · Papers with Code

MoViDNN: A Mobile Platform for Evaluating Video Quality Enhancement with Deep Neural Networks

Deep neural network (DNN) based approaches have been intensively studied to improve video quality thanks to their fast advancement in recent years. These approaches are designed mainly for desktop devices due to their high computational cost. However, with the increasing performance of mobile devices in recent years, it became possible to execute DNN based approaches in mobile devices. Despite having the required computational power, utilizing DNNs to improve the video quality for mobile devices is still an active research area. In this paper, we propose an open-source mobile platform, namely MoViDNN, to evaluate DNN based video quality enhancement methods, such as super-resolution, denoising, and deblocking. Our proposed platform can be used to evaluate the DNN based approaches both objectively and subjectively. For objective evaluation, we report common metrics such as execution time, PSNR, and SSIM. For subjective evaluation, Mean Score Opinion (MOS) is reported. The proposed platform is available publicly at https://github.com/cd-athena/MoViDNN

arXiv:2201.04358공식 코드 · Papers with Code

Coarse-to-Fine Embedded PatchMatch and Multi-Scale Dynamic Aggregation for Reference-based Super-Resolution

Reference-based super-resolution (RefSR) has made significant progress in producing realistic textures using an external reference (Ref) image. However, existing RefSR methods obtain high-quality correspondence matchings consuming quadratic computation resources with respect to the input size, limiting its application. Moreover, these approaches usually suffer from scale misalignments between the low-resolution (LR) image and Ref image. In this paper, we propose an Accelerated Multi-Scale Aggregation network (AMSA) for Reference-based Super-Resolution, including Coarse-to-Fine Embedded PatchMatch (CFE-PatchMatch) and Multi-Scale Dynamic Aggregation (MSDA) module. To improve matching efficiency, we design a novel Embedded PatchMacth scheme with random samples propagation, which involves end-to-end training with asymptotic linear computational cost to the input size. To further reduce computational cost and speed up convergence, we apply the coarse-to-fine strategy on Embedded PatchMacth constituting CFE-PatchMatch. To fully leverage reference information across multiple scales and enhance robustness to scale misalignment, we develop the MSDA module consisting of Dynamic Aggregation and Multi-Scale Aggregation. The Dynamic Aggregation corrects minor scale misalignment by dynamically aggregating features, and the Multi-Scale Aggregation brings robustness to large scale misalignment by fusing multi-scale information. Experimental results show that the proposed AMSA achieves superior performance over state-of-the-art approaches on both quantitative and qualitative evaluations.

arXiv:2201.03992공식 코드 · Papers with Code

Image quality measurements and denoising using Fourier Ring Correlations

Image quality is a nebulous concept with different meanings to different people. To quantify image quality a relative difference is typically calculated between a corrupted image and a ground truth image. But what metric should we use for measuring this difference? Ideally, the metric should perform well for both natural and scientific images. The structural similarity index (SSIM) is a good measure for how humans perceive image similarities, but is not sensitive to differences that are scientifically meaningful in microscopy. In electron and super-resolution microscopy, the Fourier Ring Correlation (FRC) is often used, but is little known outside of these fields. Here we show that the FRC can equally well be applied to natural images, e.g. the Google Open Images dataset. We then define a loss function based on the FRC, show that it is analytically differentiable, and use it to train a U-net for denoising of images. This FRC-based loss function allows the network to train faster and achieve similar or better results than when using L1- or L2- based losses. We also investigate the properties and limitations of neural network denoising with the FRC analysis.

arXiv:2201.03794공식 코드 · Papers with Code

Efficient Non-Local Contrastive Attention for Image Super-Resolution

Non-Local Attention (NLA) brings significant improvement for Single Image Super-Resolution (SISR) by leveraging intrinsic feature correlation in natural images. However, NLA gives noisy information large weights and consumes quadratic computation resources with respect to the input size, limiting its performance and application. In this paper, we propose a novel Efficient Non-Local Contrastive Attention (ENLCA) to perform long-range visual modeling and leverage more relevant non-local features. Specifically, ENLCA consists of two parts, Efficient Non-Local Attention (ENLA) and Sparse Aggregation. ENLA adopts the kernel method to approximate exponential function and obtains linear computation complexity. For Sparse Aggregation, we multiply inputs by an amplification factor to focus on informative features, yet the variance of approximation increases exponentially. Therefore, contrastive learning is applied to further separate relevant and irrelevant features. To demonstrate the effectiveness of ENLCA, we build an architecture called Efficient Non-Local Contrastive Network (ENLCN) by adding a few of our modules in a simple backbone. Extensive experimental results show that ENLCN reaches superior performance over state-of-the-art approaches on both quantitative and qualitative evaluations.

arXiv:2201.01014공식 코드 · Papers with Code

Local Motion and Contrast Priors Driven Deep Network for Infrared Small Target Super-Resolution

Infrared small target super-resolution (SR) aims to recover reliable and detailed high-resolution image with high-contrast targets from its low-resolution counterparts. Since the infrared small target lacks color and fine structure information, it is significant to exploit the supplementary information among sequence images to enhance the target. In this paper, we propose the first infrared small target SR method named local motion and contrast prior driven deep network (MoCoPnet) to integrate the domain knowledge of infrared small target into deep network, which can mitigate the intrinsic feature scarcity of infrared small targets. Specifically, motivated by the local motion prior in the spatio-temporal dimension, we propose a local spatio-temporal attention module to perform implicit frame alignment and incorporate the local spatio-temporal information to enhance the local features (especially for small targets). Motivated by the local contrast prior in the spatial dimension, we propose a central difference residual group to incorporate the central difference convolution into the feature extraction backbone, which can achieve center-oriented gradient-aware feature extraction to further improve the target contrast. Extensive experiments have demonstrated that our method can recover accurate spatial dependency and improve the target contrast. Comparative results show that MoCoPnet can outperform the state-of-the-art video SR and single image SR methods in terms of both SR performance and target enhancement. Based on the SR results, we further investigate the influence of SR on infrared small target detection and the experimental results demonstrate that MoCoPnet promotes the detection performance. The code is available at https://github.com/XinyiYing/MoCoPnet.

arXiv:2112.15386공식 코드 · Papers with Code

Efficient Single Image Super-Resolution Using Dual Path Connections with Multiple Scale Learning

Deep convolutional neural networks have been demonstrated to be effective for SISR in recent years. On the one hand, residual connections and dense connections have been used widely to ease forward information and backward gradient flows to boost performance. However, current methods use residual connections and dense connections separately in most network layers in a sub-optimal way. On the other hand, although various networks and methods have been designed to improve computation efficiency, save parameters, or utilize training data of multiple scale factors for each other to boost performance, it either do super-resolution in HR space to have a high computation cost or can not share parameters between models of different scale factors to save parameters and inference time. To tackle these challenges, we propose an efficient single image super-resolution network using dual path connections with multiple scale learning named as EMSRDPN. By introducing dual path connections inspired by Dual Path Networks into EMSRDPN, it uses residual connections and dense connections in an integrated way in most network layers. Dual path connections have the benefits of both reusing common features of residual connections and exploring new features of dense connections to learn a good representation for SISR. To utilize the feature correlation of multiple scale factors, EMSRDPN shares all network units in LR space between different scale factors to learn shared features and only uses a separate reconstruction unit for each scale factor, which can utilize training data of multiple scale factors to help each other to boost performance, meanwhile which can save parameters and support shared inference for multiple scale factors to improve efficiency. Experiments show EMSRDPN achieves better performance and comparable or even better parameter and inference efficiency over SOTA methods.

arXiv:2201.00346공식 코드 · Papers with Code

Detail-Preserving Transformer for Light Field Image Super-Resolution

Recently, numerous algorithms have been developed to tackle the problem of light field super-resolution (LFSR), i.e., super-resolving low-resolution light fields to gain high-resolution views. Despite delivering encouraging results, these approaches are all convolution-based, and are naturally weak in global relation modeling of sub-aperture images necessarily to characterize the inherent structure of light fields. In this paper, we put forth a novel formulation built upon Transformers, by treating LFSR as a sequence-to-sequence reconstruction task. In particular, our model regards sub-aperture images of each vertical or horizontal angular view as a sequence, and establishes long-range geometric dependencies within each sequence via a spatial-angular locally-enhanced self-attention layer, which maintains the locality of each sub-aperture image as well. Additionally, to better recover image details, we propose a detail-preserving Transformer (termed as DPT), by leveraging gradient maps of light field to guide the sequence learning. DPT consists of two branches, with each associated with a Transformer for learning from an original or gradient image sequence. The two branches are finally fused to obtain comprehensive feature representations for reconstruction. Evaluations are conducted on a number of light field datasets, including real-world scenes and synthetic data. The proposed method achieves superior performance comparing with other state-of-the-art schemes. Our code is publicly available at: https://github.com/BITszwang/DPT.

arXiv:2112.14608공식 코드 · Papers with Code

HPRN: Holistic Prior-embedded Relation Network for Spectral Super-Resolution

Spectral super-resolution (SSR) refers to the hyperspectral image (HSI) recovery from an RGB counterpart. Due to the one-to-many nature of the SSR problem, a single RGB image can be reprojected to many HSIs. The key to tackle this ill-posed problem is to plug into multi-source prior information such as the natural spatial context-prior of RGB images, deep feature-prior or inherent statistical-prior of HSIs, etc., so as to effectively alleviate the degree of ill-posedness. However, most current approaches only consider the general and limited priors in their customized convolutional neural networks (CNNs), which leads to the inability to guarantee the confidence and fidelity of reconstructed spectra. In this paper, we propose a novel holistic prior-embedded relation network (HPRN) to integrate comprehensive priors to regularize and optimize the solution space of SSR. Basically, the core framework is delicately assembled by several multi-residual relation blocks (MRBs) that fully facilitate the transmission and utilization of the low-frequency content prior of RGBs. Innovatively, the semantic prior of RGB inputs is introduced to mark category attributes, and a semantic-driven spatial relation module (SSRM) is invented to perform the feature aggregation of clustered similar range for refining recovered characteristics. Additionally, we develop a transformer-based channel relation module (TCRM), which breaks the habit of employing scalars as the descriptors of channel-wise relations in the previous deep feature-prior, and replaces them with certain vectors to make the mapping function more robust and smoother. In order to maintain the mathematical correlation and spectral consistency between hyperspectral bands, the second-order prior constraints (SOPC) are incorporated into the loss function to guide the HSI reconstruction.

arXiv:2112.10175공식 코드 · Papers with Code

On Efficient Transformer-Based Image Pre-training for Low-Level Vision

Pre-training has marked numerous state of the arts in high-level computer vision, while few attempts have ever been made to investigate how pre-training acts in image processing systems. In this paper, we tailor transformer-based pre-training regimes that boost various low-level tasks. To comprehensively diagnose the influence of pre-training, we design a whole set of principled evaluation tools that uncover its effects on internal representations. The observations demonstrate that pre-training plays strikingly different roles in low-level tasks. For example, pre-training introduces more local information to higher layers in super-resolution (SR), yielding significant performance gains, while pre-training hardly affects internal feature representations in denoising, resulting in limited gains. Further, we explore different methods of pre-training, revealing that multi-related-task pre-training is more effective and data-efficient than other alternatives. Finally, we extend our study to varying data scales and model sizes, as well as comparisons between transformers and CNNs-based architectures. Based on the study, we successfully develop state-of-the-art models for multiple low-level tasks. Code is released at https://github.com/fenglinglwb/EDT.

arXiv:2112.09532공식 코드 · Papers with Code

Pixel Distillation: A New Knowledge Distillation Scheme for Low-Resolution Image Recognition

Previous knowledge distillation (KD) methods mostly focus on compressing network architectures, which is not thorough enough in deployment as some costs like transmission bandwidth and imaging equipment are related to the image size. Therefore, we propose Pixel Distillation that extends knowledge distillation into the input level while simultaneously breaking architecture constraints. Such a scheme can achieve flexible cost control for deployment, as it allows the system to adjust both network architecture and image quality according to the overall requirement of resources. Specifically, we first propose an input spatial representation distillation (ISRD) mechanism to transfer spatial knowledge from large images to student's input module, which can facilitate stable knowledge transfer between CNN and ViT. Then, a Teacher-Assistant-Student (TAS) framework is further established to disentangle pixel distillation into the model compression stage and input compression stage, which significantly reduces the overall complexity of pixel distillation and the difficulty of distilling intermediate knowledge. Finally, we adapt pixel distillation to object detection via an aligned feature for preservation (AFP) strategy for TAS, which aligns output dimensions of detectors at each stage by manipulating features and anchors of the assistant. Comprehensive experiments on image classification and object detection demonstrate the effectiveness of our method. Code is available at https://github.com/gyguo/PixelDistillation.

arXiv:2112.10046공식 코드 · Papers with Code

A-ESRGAN: Training Real-World Blind Super-Resolution with Attention U-Net Discriminators

Blind image super-resolution(SR) is a long-standing task in CV that aims to restore low-resolution images suffering from unknown and complex distortions. Recent work has largely focused on adopting more complicated degradation models to emulate real-world degradations. The resulting models have made breakthroughs in perceptual loss and yield perceptually convincing results. However, the limitation brought by current generative adversarial network structures is still significant: treating pixels equally leads to the ignorance of the image's structural features, and results in performance drawbacks such as twisted lines and background over-sharpening or blurring. In this paper, we present A-ESRGAN, a GAN model for blind SR tasks featuring an attention U-Net based, multi-scale discriminator that can be seamlessly integrated with other generators. To our knowledge, this is the first work to introduce attention U-Net structure as the discriminator of GAN to solve blind SR problems. And the paper also gives an interpretation for the mechanism behind multi-scale attention U-Net that brings performance breakthrough to the model. Through comparison experiments with prior works, our model presents state-of-the-art level performance on the non-reference natural image quality evaluator metric. And our ablation studies have shown that with our discriminator, the RRDB based generator can leverage the structural features of an image in multiple scales, and consequently yields more perceptually realistic high-resolution images compared to prior works.

arXiv:2112.08950공식 코드 · Papers with Code

Stable Long-Term Recurrent Video Super-Resolution

Recurrent models have gained popularity in deep learning (DL) based video super-resolution (VSR), due to their increased computational efficiency, temporal receptive field and temporal consistency compared to sliding-window based models. However, when inferring on long video sequences presenting low motion (i.e. in which some parts of the scene barely move), recurrent models diverge through recurrent processing, generating high frequency artifacts. To the best of our knowledge, no study about VSR pointed out this instability problem, which can be critical for some real-world applications. Video surveillance is a typical example where such artifacts would occur, as both the camera and the scene stay static for a long time. In this work, we expose instabilities of existing recurrent VSR networks on long sequences with low motion. We demonstrate it on a new long sequence dataset Quasi-Static Video Set, that we have created. Finally, we introduce a new framework of recurrent VSR networks that is both stable and competitive, based on Lipschitz stability theory. We propose a new recurrent VSR network, coined Middle Recurrent Video Super-Resolution (MRVSR), based on this framework. We empirically show its competitive performance on long sequences with low motion.