Unsupervised video consistent component segmentation method based on deep convolutional network

A deep convolution, unsupervised technology, applied in computer components, neural learning methods, biological neural network models, etc., can solve the problems of blurred segmentation, insufficient optical flow to represent its transformation, and global optical flow cannot be processed at the same time. , to achieve the effect of ensuring correctness

Pending Publication Date: 2021-12-03
BEIJING FILM ACAD
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] This method uses the overall optical flow information to deform the input original image when reconstructing the image, making it as close as possible to the target image. However, for two images with large differences, the overall optical flow is not enough to represent the transformation. For example, when there is overlap between components, the global optical flow cannot be processed at the same time; in addition, for the boundary of each component, the use of global optical flow processing can easily lead to blurred segmentation

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
  • Unsupervised video consistent component segmentation method based on deep convolutional network
  • Unsupervised video consistent component segmentation method based on deep convolutional network
  • Unsupervised video consistent component segmentation method based on deep convolutional network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0022] The present invention will be described in detail below with reference to the accompanying drawings and examples.

[0023] The present invention provides a method for unsupervised video consistent component segmentation based on deep convolutional network, the flow chart is as follows figure 1 shown. Firstly, a deep convolutional neural network is constructed, and the dual process of part segmentation and part assembly is introduced to form a closed loop to realize self-supervision, which can handle more complex movements.

[0024] figure 2 It is the basic framework of the deep convolutional neural network constructed by the present invention, and realizes self-supervision by introducing the dual process of component segmentation and component assembly to form a closed loop. The deep convolutional neural network of the present invention consists of three main parts, namely the image encoder ε, the segmentation decoder D and the assembly module. The image encoder ε i...

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 invention discloses an unsupervised video consistent component segmentation method based on a deep convolutional network. According to the invention, the action in the same video is subjected to different transformation at a group of same standard positions, and the picture is reconstructed by introducing component segmentation and component assembly, so that more complex motion can be processed. And the double processes of component segmentation and component assembly are closed loops, so that self-supervision can be realized, and the correctness of the segmentation result can be ensured.

Description

technical field [0001] The invention relates to the technical field of computer graphics, in particular to an unsupervised video consistent component segmentation method based on a deep convolutional network. Background technique [0002] In the field of computer graphics, video consistent part segmentation refers to dividing the image into several disjoint parts according to the characteristics of the image in the video, so that these parts show consistency or similarity in the same part. When the existing technology performs consistent part segmentation in unsupervised video, it generally performs segmentation according to the picture features of the image and the optical flow motion mode. [0003] Existing unsupervised methods for consistent segmentation (Aliaksandr Siarohin et al. Motion-supervised Co-Part Segmentation. 2020 25th International Conference on Pattern Recognition (ICPR)) use neural networks to segment consistent parts of videos. This method proposes an uns...

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): G06K9/34G06K9/62G06N3/04G06N3/08
CPCG06N3/088G06N3/047G06N3/045G06F18/214
Inventor 高庆哲王滨刘利斌陈宝权
Owner BEIJING FILM ACAD
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