Video transcoding method, device and system

A video transcoding and video technology, applied in the field of video processing, can solve the problems of increasing the number of frames from pictures to video images, not supporting video input and output, and unable to use video, and reducing deployment costs.

Active Publication Date: 2017-10-03
ALIBABA GRP HLDG LTD
View PDF5 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing machine learning in image processing is only limited to the stage of single image processing, and cannot be used in video
The reason for the analysis is that the existing machine learning framework itself does not support video input and output, and the complexity of the processing flow from pictures to video images is not just a simple increase in the number of frames

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 more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Video transcoding method, device and system
  • Video transcoding method, device and system
  • Video transcoding method, device and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0053] refer to Figure 4 , which shows a flow chart of the steps of an embodiment of a video transcoding method of the present application, which may specifically include the following steps:

[0054] Step 402, receiving the model file issued by the machine learning framework;

[0055] Preferably, the machine learning framework can use any one of Theano library, Caffe library or Torch library to train samples so as to generate template files.

[0056] It should be noted here that the model file can be obtained through training with various existing machine learning frameworks, and the embodiment of the present application does not limit the specific machine learning framework used.

[0057] In addition, the model file is obtained by offline training of the machine learning framework, which is completely independent from the actual video processing process. In this way, the template file can be generated through the most suitable machine learning framework according to differ...

Embodiment 2

[0072] refer to Figure 5 , showing a flow chart of the steps of another video transcoding method embodiment of the present application, which may specifically include the following steps:

[0073] Step 500, train the training samples through the machine learning framework to obtain the model file.

[0074] Preferably, the step of obtaining the model file by training the training sample through the machine learning framework may further include:

[0075] S 510. Collect training samples based on video processing requirements;

[0076] Preferably, it is based on the user's specific requirements for video enhancement, such as super-resolution, denoising, and the like. The training samples may be a large number of pictures collected, such as tens of thousands of pictures.

[0077] S 520. Perform offline training on the training samples to obtain a model file.

[0078] Preferably, the machine learning framework can use any one of Theano library, Caffe library or Torch library t...

Embodiment 3

[0097] In the following, a specific application of the embodiment of the present application will be described in conjunction with an actual business scenario.

[0098] Realize the function of converting video to high-definition service on MTS. In specific video processing, for example, if a user wants to remake a cartoon in high-definition, he can put forward the demand for enhancement of the cartoon video through MTS online, for example, to The video is super-resolutioned, that is, the cartoon is converted into a high-definition video.

[0099] After collecting tens of thousands of pictures based on user needs, the machine learning framework runs a specific algorithm to perform offline training on tens of thousands of pictures to obtain corresponding model files. Since the training process is completely independent from the actual video enhancement process, the training machine can be directly deployed in the experimental environment.

[0100] In the specific application of...

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

No PUM Login to view more

Abstract

The embodiment of the invention provides a video transcoding method, device and system. The method comprises the steps of receiving a model file issued by a machine learning frame; converting the model file into a simple file capable of being identified by a filter; setting a filter parameter according to the simple file; and processing a video based on the filter parameter. According to the embodiment of the invention, additional storage consumption is not increased in a video transcoding process, the system deployment cost does not need to be increased additionally, and the system resource consumption is reduced as much as possible.

Description

technical field [0001] The present application relates to the technical field of video processing, in particular to a video transcoding method, device and system. Background technique [0002] With the rapid development of Internet technology, people's demand for Internet video services is also increasing. Video transcoding is the basis of almost all Internet video services. However, due to reasons such as video shooting, copyright, and age, some videos need further processing, such as denoising, super-resolution, etc., in order to obtain better viewing quality. Therefore, after years of development, conventional video image processing techniques have been applied. [0003] Due to the rise of machine learning algorithms in recent years, it has been theoretically proved that better processing results can be achieved by using machine learning, especially complex processing of images. However, the existing machine learning in image processing is only limited to the stage of ...

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
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): H04N21/2343H04N21/4402
CPCH04N21/234309H04N21/440218H04N19/40H04N19/117G06N20/00
Inventor 徐浩晖梅大为周昌
Owner ALIBABA GRP HLDG LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap