Image Super-Resolution Using Dense Skip Connections

By Tong Tong, Gen Li, Xiejie Liu, Qinquan Gao. 2017

1,272 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Why Should We Add Early Exits to Neural Networks? (2020, background): “o, among others, face recognition [66], object detection [15], image segmentation [16], pose estimation [27],saliencydetection[43],metriclearning[76],visualattentionprediction[74],andsuper resolution [36,70]. JT approaches are also popular in the case of recurrent neural networks for sequence classi”
Progressive Fusion Video Super-Resolution Network via Exploiting Non-Local Spatio-Temporal Correlations (2019, methodology): “[4, 5], a lot of CNN-based SR methods [8, 24, 27, 33, 40] have been proposed.”
Simultaneous Enhancement and Super-Resolution of Underwater Imagery for Improved Visual Perception (2020, methodology): “Researchers have also exploited contemporary techniques [39, 40, 61] such as gradient clipping, dense skip connection, and sub-pixel convolution to improve SISR performance on standard datasets.”
Creating High Resolution Images with a Latent Adversarial Generator (2020, background): “Indeed, newer and better architectures are constantly appearing in the literature [17, 19] and LAG should be adaptable to these other architecture.”
Densely Residual Laplacian Super-Resolution (2019, methodology): “To upsample the features, SRDenseNet [14] orders the blocks sequentially followed by deconvolutional layers at the end of the network.”
To learn image super-resolution, use a GAN to learn how to do image degradation first (2018, neutral): “More advanced deep architectures for super-resolution including recursive, laplacian and dense networks have been recently proposed in [10–12].”
Learning Enriched Features for Real Image Restoration and Enhancement (2020, background): “ature fusion (SKFF) in order to consolidate the highresolution features with the help of low-resolution features, and vice versa. ing [55,40,3], progressive reconstruction [104,58], dense connections [98,103,125], attention mechanisms [123,22,124], multi-branch learning [58,65,21,62], and generative”
Learning Enriched Features for Fast Image Restoration and Enhancement (2022, neutral): “[65], [66], dense connections [7], [67], [68], attention mechanisms [69], [70], [71], multi-branch learning [66], [72], [73], [74], and generative adversarial networks (GANs) [68], [75], [76], [77].”
Learning to Maintain Natural Image Statistics (2018, background): “The first group relies on the L2 or L1 losses [10,34,35,36], which lead to high PSNR and SSIM [37] at the price of low perceptual quality, as was recently shown in [30,38].”
Deep Video Super-Resolution Network Using Dynamic Upsampling Filters Without Explicit Motion Compensation (2018, methodology): “With the success of deep learning in computer vision as in image classification [19] and image segmentation [25], the deep learning based single-image super-resolution (SISR) methods have emerged [3, 17, 31, 21, 33, 20, 35, 30].”
Learned Image Downscaling for Upscaling Using Content Adaptive Resampler (2019, methodology): “To further validate the effectiveness of the CAR image downscaling model, we evaluated the CAR model trained with another four state-of-the-art deep SR models, i . e ., the SRDenseNet [34], D-DBPN [52], RDN [35] and RCAN [37], using 4 × downscaling and upscaling factor on ﬁve testing datasets.”
Residual Dense Network for Image Super-Resolution (2018, methodology): “When compared with persistent CNN models ( SRDenseNet [30] and MemNet [25]), our RDN performs the best on all datasets with all scaling factors.”
Unsupervised Learning for Real-World Super-Resolution (2019, neutral): “Therefore, research in the field has long relied on the use of known degradation operators such as bicubic kernel in order to artificially generate a corresponding LR image [9, 31, 34].”
GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing (2019, methodology): “Selection of pre-processing method: Pre-processing is widely used in image preparation to facilitate follow-up operations [39, 27].”
2018 PIRM Challenge on Perceptual Image Super-resolution (2018, methodology): “While the first attempt to apply a conv-net to image SR [6] did not significantly surpass the performance of prior methods, it set the ground for major improvements in PSNR/SSIM values over the course of the several following years [15, 17, 18, 19, 39, 11, 52, 34, 10, 51].”
Edge-Enhanced GAN for Remote Sensing Image Superresolution (2019, methodology): “Different from the traditional dense block [18], [51], the subsequent layer in our work can share and fuse the feature maps extracted from multiple preceding con-”
Image Super-Resolution via Dual-State Recurrent Networks (2018, methodology): “Specifically, convolutional neural networks (CNNs) and residual learning [16] have been widely applied in much recent SR work [10, 12, 20, 21, 22, 25, 35, 38].”
Deep Learning for Image Super-Resolution: A Survey (2019, methodology): “Since the transposed convolution enlarges the image size in an end-to-end manner while maintaining a connectivity pattern compatible with vanilla convolution, it is widely used as upsampling layers in SR models [57], [78], [79], [85].”