18379519. IMAGE SUPER-RESOLUTION NEURAL NETWORKS simplified abstract (GOOGLE LLC)
Contents
- 1 IMAGE SUPER-RESOLUTION NEURAL NETWORKS
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 IMAGE SUPER-RESOLUTION NEURAL NETWORKS - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Unanswered Questions
- 1.11 Original Abstract Submitted
IMAGE SUPER-RESOLUTION NEURAL NETWORKS
Organization Name
Inventor(s)
Cristina Nader Vasconcelos of Montreal (CA)
Ahmet Cengiz Oztireli of Zurich (CH)
Andrea Tagliasacchi of Toronto (CA)
Kevin Jordan Swersky of Toronto (CA)
Mark Jeffrey Matthews of Los Angeles CA (US)
Milad Olia Hashemi of San Francisco CA (US)
IMAGE SUPER-RESOLUTION NEURAL NETWORKS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18379519 titled 'IMAGE SUPER-RESOLUTION NEURAL NETWORKS
Simplified Explanation
The patent application describes methods, systems, and apparatus for processing an input image using a super-resolution neural network to generate a higher resolution version of the input image.
- Processing input image using an encoder subnetwork to generate a feature map
- Generating an updated feature map by applying convolutional filters parametrized by a hyper neural network
- Processing the updated feature map using a projection subnetwork to generate the up-sampled image
Potential Applications
The technology can be used in image processing applications such as enhancing the resolution of images captured by cameras or improving the quality of medical imaging.
Problems Solved
This technology addresses the issue of low-resolution images by using neural networks to generate higher resolution versions of the input images.
Benefits
The benefits of this technology include improved image quality, enhanced details in images, and the ability to upscale images without losing quality.
Potential Commercial Applications
Commercial applications of this technology could include software for enhancing image quality in photography, medical imaging devices, and video processing tools.
Possible Prior Art
One possible prior art for this technology could be traditional image upscaling techniques that do not utilize neural networks for generating high-resolution images.
Unanswered Questions
How does this technology compare to traditional image upscaling methods?
This article does not provide a direct comparison between this technology and traditional image upscaling methods.
What are the computational requirements for implementing this super-resolution neural network?
The article does not delve into the computational requirements needed to implement this super-resolution neural network.
Original Abstract Submitted
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for processing an input image using a super-resolution neural network to generate an up-sampled image that is a higher resolution version of the input image. In one aspect, a method comprises: processing the input image using an encoder subnetwork of the super-resolution neural network to generate a feature map; generating an updated feature map, comprising, for each spatial position in the updated feature map: applying a convolutional filter to the feature map to generate a plurality of features corresponding to the spatial position in the updated feature map, wherein the convolutional filter is parametrized by a set of convolutional filter parameters that are generated by processing data representing the spatial position using a hyper neural network; and processing the updated feature map using a projection subnetwork of the super-resolution neural network to generate the up-sampled image.
