2020
DOI: 10.1007/978-3-030-58574-7_6
|View full text |Cite
|
Sign up to set email alerts
|

Journey Towards Tiny Perceptual Super-Resolution

Abstract: Please refer to published version for the most recent bibliographic citation information. If a published version is known of, the repository item page linked to above, will contain details on accessing it.

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
23
0

Year Published

2021
2021
2022
2022

Publication Types

Select...
5
3
1

Relationship

2
7

Authors

Journals

citations
Cited by 46 publications
(23 citation statements)
references
References 46 publications
0
23
0
Order By: Relevance
“…Guo et al [9] proposes to search for cell structures and upsampling positions with reinforcement learning. Recently, TPSR [19] has adopted reinforcement learning to find an efficient GAN-based SR model, resulting a tiny SR model that performes well on both perceptual and distortion metrics. However, most of the prior SR methods with NAS utilized reinforcement learning or evolutionary methods that were time-consuming.…”
Section: Sisr With Neural Architecture Searchmentioning
confidence: 99%
See 1 more Smart Citation
“…Guo et al [9] proposes to search for cell structures and upsampling positions with reinforcement learning. Recently, TPSR [19] has adopted reinforcement learning to find an efficient GAN-based SR model, resulting a tiny SR model that performes well on both perceptual and distortion metrics. However, most of the prior SR methods with NAS utilized reinforcement learning or evolutionary methods that were time-consuming.…”
Section: Sisr With Neural Architecture Searchmentioning
confidence: 99%
“…It is difficult to deploy these models to the equipment with low computing power. For real-world applications, lightweight and efficient SR models have also been designed in recent years, including handcrafted SR neural networks [16,16,2] and neural architecture search (NAS) based SR methods [5,32,19,27].…”
Section: Introductionmentioning
confidence: 99%
“…Super-resolution quantization technology [8,[28][29][30][31][32] has recently started to do a lot of work. At the beginning, super-resolution reconstruction focused on the results obtained with the accuracy of each pixel.…”
Section: Super-resolution Model In Quantizationmentioning
confidence: 99%
“…Since then, a number of alternative approaches have been proposed in order to reduce the significant searching time introduced by the need to train each proposed model. Among those, Pham et al proposed to train models using only one epoch but preserve weights between trainings [19] -an approach classified as one-shot method that is very close to the differentiable NAS described below [20,21]. In general, reduced training (with or without weight sharing) can be found in many works using NAS [22].…”
Section: Related Workmentioning
confidence: 99%