Analyzing Perception-Distortion Tradeoff Using Enhanced Perceptual Super-Resolution Network
Citations
837 citations
Cites background from "Analyzing Perception-Distortion Tra..."
...In practice, researchers often combine multiple loss functions by weighted average [8], [25], [27], [46], [141] for constraining different aspects of the generation process, especially for distortion-perception tradeoff [25], [103], [142], [143], [144]....
[...]
428 citations
205 citations
Cites background from "Analyzing Perception-Distortion Tra..."
...face frontal view generation [50], generating new human poses [51], photos to emojis [52], photograph editing [53]–[55], face aging [56], [57], photo blending [58], super resolution [59]–[61], photo inpainting [62]–[64], video prediction [65], and 3D object generation [66], [67]; among...
[...]
113 citations
88 citations
Cites background from "Analyzing Perception-Distortion Tra..."
...Generation of realistic images has wide range of practical applications, such as anime character generation [159]–[164], image synthesis [165]–[168], super resolution [10], [124], [169]–[177], image editing and blending [178], [179], inpainting [125], [180]–[182], interactive image generation [183], [184], human pose estimation [185], [186], face aging [187], [188], 3D object detection [189]–[192], etc....
[...]
References
1,923 citations
"Analyzing Perception-Distortion Tra..." refers background in this paper
...establish a complex mapping between LR and HR image pairs. The works in [16,15] were some of the early approaches to learn such a complex mapping using example-pairs of LR and HR training patches. In [18], the presence of patch redundancies across scales within an image was exploited to generate more realistic textures. This idea was further extended by [21] wherein self-dictionaries were constructed ...
[...]
1,882 citations
1,647 citations
"Analyzing Perception-Distortion Tra..." refers background in this paper
...The works in [16,15] were some of the early approaches to learn such a complex mapping using example-pairs of LR and HR training patches....
[...]
1,565 citations
"Analyzing Perception-Distortion Tra..." refers methods in this paper
... SISR with deep networks gained momentum with the primal work of Chao et al. [12]. While [12] used a 3 layer convolutional neural network (CNN), the subsequent works used deeper network architectures [23,24] and new techniques to improve the restoration accuracy [31,20] and computational complexity [40,13]. Despite signicant progress in both reconstruction accuracy and speed, a majority of the existing ...
[...]
...ared in [11,12] (SRCNN) wherein a 3 layer network was employed to learn the mapping between the desired HR image and its bicubic up-sampled LR image. This was followed by deeper network architectures [23,24] promising 4 Subeesh Vasu, Nimisha T. M., Rajagopalan A.N. performance improvement over SRCNN. [23] proposed to use residual-learning and gradient clipping with a high-learning rate, whereas [24] reli...
[...]
1,418 citations
"Analyzing Perception-Distortion Tra..." refers methods in this paper
...le dictionary. HR images from the web with similar contents were used with-in a structure-aware matching criterion to super-resolve landmark images in [48]. The class of neighbor embedding approaches [8,3,17,44,45] aim to nd similar looking LR training patches from a low dimensional manifold and then combine their corresponding HR patches for resolution enhancement. The overtting tendency of neighborhood appro...
[...]