Face optical flow estimation network training and face optical flow estimation method and device

A training method and optical flow technology, applied in the training field of face optical flow estimation network, can solve the problem of difficult to obtain the true value of face optical flow, and achieve the effect of reducing the labeling cost, high quality, and pure optical flow

Pending Publication Date: 2021-12-24
BEIJING HORIZON INFORMATION TECH CO LTD
View PDF7 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Supervised optical flow estimation algorithms based on deep learning often require a large number of true values ​​for model training, but the true value of face optical flow in real scenes is very difficult to obtain

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
  • Face optical flow estimation network training and face optical flow estimation method and device
  • Face optical flow estimation network training and face optical flow estimation method and device
  • Face optical flow estimation network training and face optical flow estimation method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0040] Hereinafter, exemplary embodiments according to the present disclosure will be described in detail with reference to the accompanying drawings. Apparently, the described embodiments are only some of the embodiments of the present disclosure, rather than all the embodiments of the present disclosure, and it should be understood that the present disclosure is not limited by the exemplary embodiments described here.

[0041] It should be noted that relative arrangements of components and steps, numerical expressions and numerical values ​​set forth in these embodiments do not limit the scope of the present disclosure unless specifically stated otherwise.

[0042] Those skilled in the art can understand that terms such as "first" and "second" in the embodiments of the present disclosure are only used to distinguish different steps, devices or modules, etc. necessary logical sequence.

[0043] It should also be understood that in the embodiments of the present disclosure, "...

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 discloses a face optical flow estimation network training and face optical flow estimation method and a device. The method comprises the following steps: inputting a first face image and a second face image corresponding to two adjacent frames of images in a video into a face optical flow estimation network, and obtaining a predicted face optical flow value corresponding to the first face image; respectively carrying out Gaussian blur processing on the first face image and the second face image to obtain a first smooth image and a second smooth image; determining the network loss of the face optical flow estimation network based on the first smooth image, the second smooth image and the predicted face optical flow value; and training of the face optical flow estimation network is supervised based on the network loss. According to the embodiment of the invention, the network loss of the face optical flow estimation network is determined based on the first smooth image, the second smooth image and the predicted face optical flow value, the face optical flow truth value does not need to be collected, the face optical flow estimation network is trained under an unsupervised condition, and the marking cost is reduced.

Description

technical field [0001] The present disclosure relates to computer vision technology, in particular to a face optical flow estimation network training method and a face optical flow estimation method and device. Background technique [0002] Face dense optical flow estimation is used to obtain the motion of each pixel in the face image between frames, and has a wide range of applications in the fields of expression / micro-expression recognition and behavior recognition. In recent years, with the development of deep learning technology, the dense optical flow estimation technology based on deep learning has surpassed the traditional method and achieved better results. Supervised optical flow estimation algorithms based on deep learning often require a large number of true values ​​for model training, but the true value of face optical flow in real scenes is very difficult to obtain Contents of the invention [0003] In order to solve the above-mentioned technical problems, t...

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): G06T5/00G06K9/00G06T3/00G06N3/04G06N3/08
CPCG06T5/002G06T3/0006G06N3/04G06N3/084G06T2207/30201G06T2207/20081
Inventor 于雷隋伟张骞
Owner BEIJING HORIZON INFORMATION TECH CO 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