TAPE: Task-Agnostic Prior Embedding for Image Restoration

Liu, Lin; Xie, Lingxi; Zhang, Xiaopeng; Yuan, Shanxin; Chen, Xiangyu; Zhou, Wengang; Li, Houqiang; Tian, Qi

doi:10.1007/978-3-031-19797-0_26

Cited by 8 publications

(2 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Image restoration aims to generate a high-quality image from a degraded image. Since deep learning was successfully applied to various image restoration tasks such as image super-resolution [25], image denoising [7], and compression artifacts reduction [6], a large number of deep networks have been proposed for image restoration [9], [19], [20], [21], [22], [65], [67], [68], [69], [70], [71], [72]. Before Transformer is applied to low-level vision tasks and demonstrates impressive performance, CNN-based networks dominates this field.…”

Section: Deep Network For Image Restorationmentioning

confidence: 99%

Magnetizing Inrush Current Elimination Strategy Based on a Parallel Type Asynchronous Closing Hybrid Transformer

Chen

Dong

et al. 2023

IEEE Trans. Power Electron.

View full text Add to dashboard Cite

Transformer-based methods have shown impressive performance in image restoration tasks, such as image super-resolution and denoising. However, we find that these networks can only utilize a limited spatial range of input information through attribution analysis. This implies that the potential of Transformer is still not fully exploited in existing networks. In order to activate more input pixels for better restoration, we propose a new Hybrid Attention Transformer (HAT). It combines both channel attention and window-based self-attention schemes, thus making use of their complementary advantages. Moreover, to better aggregate the cross-window information, we introduce an overlapping cross-attention module to enhance the interaction between neighboring window features. In the training stage, we additionally adopt a same-task pre-training strategy to further exploit the potential of the model for further improvement. Extensive experiments have demonstrated the effectiveness of the proposed modules. We further scale up the model to show that the performance of the SR task can be greatly improved. Besides, we extend HAT to more image restoration applications, including real-world image super-resolution, Gaussian image denoising and image compression artifacts reduction. Experiments on benchmark and real-world datasets demonstrate that our HAT achieves state-of-the-art performance both quantitatively and qualitatively. Codes and models are publicly available at https://github.com/XPixelGroup/HAT.

show abstract

Section: Deep Network For Image Restorationmentioning

confidence: 99%

Magnetizing Inrush Current Elimination Strategy Based on a Parallel Type Asynchronous Closing Hybrid Transformer

Chen

Dong

et al. 2023

IEEE Trans. Power Electron.

View full text Add to dashboard Cite

show abstract

“…Image restoration networks. Since SRCNN (Dong et al 2014) first introduces deep learning to image SR and obtains superior performance over conventional methods, numerous deep networks have been proposed for various image restoration tasks such as image SR (Lim et al 2017;Zhang et al 2018b;Chen et al 2023b), denoising (Zhang et al 2017a,b;Wang et al 2022b), deblurring (Abuolaim and Brown 2020;Chen et al 2021b), deraining (Yang et al 2017a;Chen et al 2021a;Liu et al 2022), dehazing (Yang et al 2017b;Tu et al 2022;Song et al 2023) Difference from the previous network design research.…”

Section: Related Workmentioning

confidence: 99%

Pricing the residential housing of Shanghai and Shenzhen

Chen¹

View full text Add to dashboard Cite

Despite the significant progress made by deep models in various image restoration tasks, existing image restoration networks still face challenges in terms of task generality. An intuitive manifestation is that networks which excel in certain tasks often fail to deliver satisfactory results in others.To illustrate this point, we select five representative image restoration networks and conduct a comparative study on five classic image restoration tasks. First, we provide a detailed explanation of the characteristics of different image restoration tasks and backbone networks. Following this, we present the benchmark results and analyze the reasons behind the performance disparity of different models across various tasks. Drawing from this comparative study, we propose that a general image restoration backbone network needs to meet the functional requirements of diverse tasks. Based on this principle, we design a new general image restoration backbone network, X-Restormer. Extensive experiments demonstrate that X-Restormer possesses good task generality and achieves state-of-the-art performance across a variety of tasks.

show abstract

Deep Learning for Screen-Shot Image Demoiréing: A Survey

Hou

Wang

et al. 2022

IEEE Access

View full text Add to dashboard Cite

Image demoiréing is an important image processing technology in computer vision, used to remove the moiré from images and improve the image quality. In recent years, the image demoiréing technique based on the deep learning method has gained more attention and achieved good results, but it still has some limitations. This paper aims to provide a review and perspective on the recent advances in deep learning-based image demoiréing techniques. First, the definition and production principle of the image moiré pattern are given. Common datasets and image quality evalution methods in demoiréing studies are analyzed. Then two internationally famous competitions in image demoiréing are introduced. Second, the research status of the supervised demoiréing technique is summarized from four dimensions: sampling method, model network design, baseline model, and training learning strategy. Recent progress made by the mainstream model of unsupervised deep learning in the field of image demoiréing is summarized. The typical application of the image demoiréing technique in panel defect detection and digital radiography is analyzed. The performance and the image quality of the above mentioned models based on different data sets are evaluated in detail. Finally, this paper analyzes and forelocks the problems to be solved in the coming years. INDEX TERMS Image demoiréing, screen-shot image, deep learning, convolutional neural networks (CNN)

show abstract

TAPE: Task-Agnostic Prior Embedding for Image Restoration

Cited by 8 publications

References 58 publications

Magnetizing Inrush Current Elimination Strategy Based on a Parallel Type Asynchronous Closing Hybrid Transformer

Magnetizing Inrush Current Elimination Strategy Based on a Parallel Type Asynchronous Closing Hybrid Transformer

Pricing the residential housing of Shanghai and Shenzhen

Deep Learning for Screen-Shot Image Demoiréing: A Survey

Contact Info

Product

Resources

About