Superdivision network model training method, device, electronic device and storage medium
A network model and training method technology, which is applied in biological neural network models, neural learning methods, character and pattern recognition, etc., can solve the problems of slow model calculation speed, low versatility, and difficulty in real-time operation, and achieve super sub-division performance, guarantee super-resolution performance, and ensure the effect of inter-frame continuity
Active Publication Date: 2021-11-19
TENCENT TECH (SHENZHEN) CO LTD
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
At present, video super-resolution based on deep learning requires training multiple video super-resolution models for different scenes, that is, customized design for specific scenes is required, and the versatility is not high; more model parameters are required, which leads to a slowdown in the calculation speed of the model. Slow, difficult to run in real time on terminal platforms with low computing power (such as embedded devices, etc.) or online
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0029] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiment of the application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiment of the application. Obviously, the described embodiment is only It is an embodiment of a part of the application, but not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by persons of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.
[0030] The terms "first", "second", "third" and "fourth" in the specification and claims of the present application and the drawings are used to distinguish different objects, rather than to describe a specific order . Furthermore, the terms "include" and "have", as well as any variations thereof, are intended to cover a non-exclusive inclusion. For ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The application provides a super-resolution network model training method, device, electronic equipment, and storage medium, which relate to the field of computer vision technology. The method includes: using the first network model to perform knowledge distillation on the second network model; and during the knowledge distillation process In, at least one of the mutual information loss, the first timing consistency loss and the second timing consistency loss is calculated; then, based on the mutual information loss, the first timing consistency loss and the second timing consistency loss at least One item, adjusting the parameters of the second super-resolution network model to obtain the trained second super-resolution network model. This application introduces knowledge distillation technology and designs supervisory indicators suitable for the field of video super-resolution, which can reduce the complexity of the super-resolution network model on the basis of ensuring the performance of the super-resolution network model, and further benefit the second super-resolution network model. The sub-network model runs in real time on a terminal platform with low computing power (such as embedded devices, etc.) or online.
Description
technical field [0001] The embodiments of the present application relate to the technical field of computer vision of artificial intelligence, and more specifically, to a super-resolution network model training method, device, electronic device, and storage medium. Background technique [0002] Video super-resolution refers to the generation of high-resolution video based on a given low-resolution video through a super-resolution algorithm. [0003] With the development of deep learning, deep learning methods in artificial intelligence have been applied to the field of video super-resolution due to their powerful and flexible feature extraction capabilities, and have gradually become the mainstream method of video super-resolution. At present, video super-resolution based on deep learning requires training multiple video super-resolution models for different scenes, that is, customized design for specific scenes is required, and the versatility is not high; more model parame...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F18/214
Inventor 谢植淮
Owner TENCENT TECH (SHENZHEN) CO LTD