The document discusses advancements in deep learning-based super-resolution algorithms, highlighting key milestones such as SRCNN, VDSR, DRRN, SRGAN, and EDSR. It includes references to challenges like the NTIRE challenge in image restoration and several resource links related to the evaluations of these models. Additionally, it mentions scalable machine learning using Kubernetes for deep learning deployment.

1. Scaling up Deep Learning Based Super
Resolution Algorithms
Xiaoyong Zhu
Microsoft Cloud AI Group

2. CNTK implementation

4. Let’s Enhance

6. Image Source

7. Image Source

8. because human vision is more sensitive to luminance (“black and white”) differences than
chromatic differences

9. A few milestones including SRCNN, VDSR, DRRN, SRGAN

10. • SRCNN (First to apply deep learning to Super Resolution, 2014)
• VDSR (Very Deep Convolutional Networks, 2015)
• DRRN (Deep Recursive Residual Network, CVPR 2017)
• SRGAN (Photo-Realistic using GANs, CVPR 2017)
• EDSR (Enhanced version using part of SRGAN’s work. Winner of
NTIRE2017 Super resolution challenge)
• NTIRE Challenge (New Trends in Image Restoration and Enhancement) is a
challenge in this area (http://www.vision.ee.ethz.ch/ntire17/)

11. Image Source
linear
bilinear bicubic

13. • Bicubic interpolation
• VDSR (Very Deep Convolutional Networks, 2015)
• DRRN (Deep Recursive Residual Network, CVPR 2017)
• SRGAN (Photo-Realistic using GANs, CVPR 2017)
• EDSR (Enhanced version using part of SRGAN’s work. Winner of
NTIRE2017 Super resolution challenge)
• NTIRE Challenge (New Trends in Image Restoration and Enhancement) is a
challenge in this area (http://www.vision.ee.ethz.ch/ntire17/)

14. http://cs231n.github.io/understanding-cnn/

15. Link to paper
Bicubic
SRCNN

17. • Bicubic interpolation
• SRCNN (First to apply deep learning to Super Resolution, 2014)
• DRRN (Deep Recursive Residual Network, CVPR 2017)
• SRGAN (Photo-Realistic using GANs, CVPR 2017)
• EDSR (Enhanced version using part of SRGAN’s work. Winner of
NTIRE2017 Super resolution challenge)
• NTIRE Challenge (New Trends in Image Restoration and Enhancement) is a
challenge in this area (http://www.vision.ee.ethz.ch/ntire17/)

18. ResNet architecture
Image frequency CNTK Code

19. Code available in CNTK

20. • Bicubic interpolation
• SRCNN (First to apply deep learning to Super Resolution, 2014)
• VDSR (Very Deep Convolutional Networks, 2015)
• SRGAN (Photo-Realistic using GANs, CVPR 2017)
• EDSR (Enhanced version using part of SRGAN’s work. Winner of
NTIRE2017 Super resolution challenge)
• NTIRE Challenge (New Trends in Image Restoration and Enhancement) is a
challenge in this area (http://www.vision.ee.ethz.ch/ntire17/)

24. The coolest idea in ML in the last twenty years - Yann Lecun

26. https://www.slideshare.net/xavigiro/deep-learning-for-computer-vision-generative-models-and-adversarial-training-upc-2016
z
G(z)
D(x)
x
D(G(z))
G
D

27. http://people.eecs.berkeley.edu/~junyanz/projects/gvm/

29. Image source: http://kvfrans.com/visualizing-features-from-a-
convolutional-neural-network/

31. https://www.cntk.ai/pythondocs/CNTK_302A_Evaluation_of_Pretrained_Su
per-resolution_Models.html
Bicubic DRRN SRGAN

32. Bicubic DRRN SRGAN
https://www.cntk.ai/pythondocs/CNTK_302A_Evaluation_of_Pretrained_Su
per-resolution_Models.html

33. SRCNN
VDSR
DRRN
SRGAN
EDSR

36. here
here

37. Scalable Machine Learning using Kubernetes
• Slides: bit.ly/DLwithK8S
• Tutorial for deploying DL with K8S using acs_engine:
bit.ly/K8SwithACSEngine
• Tutorial for deploying DL with managed K8S: aka.ms/AKS_GPU
• Azure Machine Learning simplification to K8S: aka.ms/AMLtoACS
• Batch AI for training DL at scale: bit.ly/deepbait