Unlock instant, AI-driven research and patent intelligence for your innovation.

Device and method for separating a picture into foreground and background using deep learning

A picture and equipment technology, applied in the field of separation of moving foreground objects and static background scenes, can solve problems such as difficult to segment small-sized foreground objects

Pending Publication Date: 2021-04-09
HUAWEI TECH CO LTD
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Due to size reduction, it becomes difficult to segment small-sized foreground objects

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
  • Device and method for separating a picture into foreground and background using deep learning
  • Device and method for separating a picture into foreground and background using deep learning
  • Device and method for separating a picture into foreground and background using deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0082] figure 1 A device 100 according to an embodiment of the invention is shown. The device 100 is used to separate a picture 101 into a foreground and a background, for example into a moving object and a static scene. to this end, figure 1 The device 100 is used for adopting CNN (CNN model, CNN architecture), that is, for separating pictures 101 through deep learning. The device 100 may be an image processor, a computer, a microprocessor, etc. or a plurality thereof or any combination thereof implementing a CNN.

[0083] A CNN is used to receive a picture 101 and a background model image 102 as input 101 , 102 . The background model image 102 may be an image of the scene, monitored by a surveillance camera that also provides pictures, taken beforehand (or at some definite time) without any (moving) foreground objects, or may be estimated as within a sliding window The median at each pixel location of all pictures (or frames) close to the current picture (or current fram...

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

Embodiments of the present invention relate to field of separating pictures, particularly pictures of a surveillance video, into foreground and background. A device and method are provided that employ a Convolutional Neural Network (CNN), i.e. are based on deep learning. The CNN is configured to receive as an input the picture and a background model image. The CNN is configured to generate feature maps of different resolution based on the input, wherein the resolution of feature maps is gradually reduced. Based on the feature maps, the CNN is configured to generate activation maps of different resolution, wherein the resolution of activation maps is gradually increased. Further, the CNN is configured to output a 1-channel probability map having the same resolution as the picture, wherein each pixel of the output 1-channel probability map corresponds to a pixel of the picture and indicates a probability that the corresponding pixel of the picture is associated with a foreground object or with a background object.

Description

technical field [0001] Embodiments of the present invention relate to the task of separating a picture (eg, a video, especially a picture of a surveillance video) into a foreground and a background. Specifically, it involves separating moving foreground objects from static background scenes. To this end, the present invention proposes a device and method, using a convolutional neural network (Convolutional Neural Network, CNN), that is, separation based on deep learning. Background technique [0002] On a global scale, the scale of camera networks continues to expand, resulting in massive surveillance video data. This video data requires an efficient video analytics pipeline to provide timely and accurate useful information to relevant authorities. [0003] Segmentation is a key component of traditional video analysis for extracting moving foreground objects from static background scenes. At the picture level, segmentation can be viewed as grouping the pixels of a picture...

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): G06N3/04G06N3/08G06T7/11G06T7/194
CPCG06N3/084G06T7/11G06T7/194G06T2207/20084G06N3/045
Inventor 泰·维·黄马库斯·布伦纳王洪斌唐健
Owner HUAWEI TECH CO LTD