Statements (50)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
gptkb:Cloud_Computing_Service
|
| gptkbp:application |
Image_Super-Resolution
|
| gptkbp:architect |
Three-layer CNN
|
| gptkbp:code |
Open Source
|
| gptkbp:community |
Widely adopted
|
| gptkbp:dataUsage |
Set14
BSD100 Set5 |
| gptkbp:developedBy |
Dong_et_al.
|
| gptkbp:evaluates |
Structural Similarity Index (SSIM
|
| gptkbp:featuresExhibits |
Patch-based_Feature_Extraction
|
| https://www.w3.org/2000/01/rdf-schema#label |
SRCNN
|
| gptkbp:improves |
Bicubic_Interpolation
|
| gptkbp:inputOutput |
Low-Resolution Images
33x33 patches |
| gptkbp:keyIssues |
First CNN for image super-resolution
|
| gptkbp:library |
gptkb:Keras
TensorFlow |
| gptkbp:losses |
Mean Squared Error
|
| gptkbp:notableFeature |
End-to-End Learning
|
| gptkbp:officialLanguage |
Python
|
| gptkbp:pageCount |
57,000
|
| gptkbp:performance |
Peak_Signal-to-Noise_Ratio_(PSNR)
|
| gptkbp:powerOutput |
High-Resolution Images
|
| gptkbp:predecessor |
Traditional Methods
|
| gptkbp:primaryFunction |
ReLU
|
| gptkbp:productQuality |
Improved over traditional methods
|
| gptkbp:realTimeTracking |
Limited
|
| gptkbp:relatedPatent |
gptkb:VDSR
gptkb:EDSR SRGAN DCSCN ESPCN SRResNet CARN GAN-based_Super-Resolution LapSRN RCAN SISR |
| gptkbp:researchField |
Deep Learning
Computer_Vision |
| gptkbp:researchInterest |
Influenced subsequent super-resolution methods
Published in IEEE Transactions on Pattern Analysis and Machine Intelligence |
| gptkbp:screenSize |
High-resolution images
|
| gptkbp:successor |
FSRCNN
|
| gptkbp:theory |
High
|
| gptkbp:training |
gptkb:ImageNet
End-to-End Training Hours to days |
| gptkbp:yearEstablished |
2014
|