Is it really possible to zoom and enhance images like in the CSI movies? Let's find out how image super-resolution works in the real world.
References
Image Super-Resolution Using Deep Convolutional Networks (SRCNN Paper)
[ Ссылка ]
DIV2K dataset: DIVerse 2K resolution high quality images
[ Ссылка ]
Mean Squared Error: Love It or Leave It?
[ Ссылка ]
Al Bovik Gives Primetime Emmy Award Acceptance Speech
[ Ссылка ]
Image quality assessment: from error visibility to structural similarity (SSIM Paper)
[ Ссылка ]
Loss Functions for Image Restoration with Neural Networks
[ Ссылка ]
Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network (SRGAN Paper)
[ Ссылка ]
ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks
[ Ссылка ]
ESRGAN GitHub Repository
[ Ссылка ]
AI Neural Networks being used to generate HQ textures for older games
[ Ссылка ]
Deep Network Interpolation for Continuous Imagery Effect Transition
[ Ссылка ]
Zoom to Learn, Learn to Zoom
[ Ссылка ]
See Better and Further with Super Res Zoom on the Pixel 3
[ Ссылка ]
Google Pixel Super Res Zoom
[ Ссылка ]
Handheld Multi-Frame Super-Resolution
[ Ссылка ]
Pixel Recursive Super Resolution
[ Ссылка ]
Deep Learning Crash Course References
Generative Adversarial Networks
[ Ссылка ]
Optimization Tricks: momentum, batch-norm, and more
[ Ссылка ]
How Super Resolution Works
Теги
super-resolutionsuperresolutionenhance super resolutionsuper resolution aimachine learning super resolutionmulti frame image resolutionsingle image super resolutionimage enhancementimage processingdeep learningmachine learningartificial intelligenceimage analysisneural networksaiimage enhancingenhanceimage upsamplingimage upscalingcsi zoomsuper resolutionsuperresolution gansuper-resolution from a single imageesrgangigapixel
