|
SRTGAN: Triplet Loss Based Generative Adversarial Network for Real-World Super-ResolutionEasyChair Preprint 7807, version historyVersion | Date | Pages | Version notes |
---|
1 | April 18, 2022 | 9 | | 2 | October 24, 2022 | 15 | Camera-Ready version of the paper accepted at the Computer Vision and Image Processing Conference (CVIP) 2022. The revised version includes minor changes such as: - paper formatted to 15 pages (according to conference format)
- Acronyms for PSNR, LPIPS, SSIM, etc. are defined.
- Fig. 2 caption changed to training framework rather than proposed architecture for better understanding.
- Improved the sentence "The first GAN-based framework
called SRGAN [14] introduced.. " such that it was the first to introduce perceptual loss, in addition to using the content and adversarial loss.
|
Keyphrases: Generative Adversarial Networks, deep learning, image super-resolution |
|
|