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

Methods and image processing entities for applying convolutional neural networks to images

A technology of convolutional neural network and image processing device, which is applied in image data processing, neural learning method, biological neural network model, etc.

Active Publication Date: 2021-11-16
AXIS
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] The problem associated with the above convolutional neural networks is that applying a fully connected layer at every possible offset is computationally demanding

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
  • Methods and image processing entities for applying convolutional neural networks to images
  • Methods and image processing entities for applying convolutional neural networks to images
  • Methods and image processing entities for applying convolutional neural networks to images

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] Throughout the following description, like reference numerals are used to refer to like acts, modules, circuits, components, items, elements, units etc., where applicable. In this specification, "feature detection" refers to the recognition, classification or detection of objects.

[0026] For a better understanding of the advantages of the embodiments herein, refer to figure 1 Provides a brief description of known convolutional neural networks for feature detection in images. exist figure 1 In , the feature detection phase refers to the recognition, classification, or detection of features. In this context, a "feature" may correspond to a cat, a car, or any other object or part of an object that may be shown in the image.

[0027] figure 1 A neural network 101 is shown comprising a convolutional stage 102 and a feature detection stage 103 which may sometimes be referred to as a fully connected stage.

[0028] In action A1 , the convolution stage 102 convolves the ...

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

A method and image processing entity (400) for applying a convolutional neural network to an image are disclosed. The image processing entity (400) processes (A020) the image while providing the feature map using the convolution kernel, whereby the second feature map size of the feature map is larger than the first feature map size of the feature map with which the feature kernel was trained. Furthermore, the image processing entity (400) repeatedly applies (A040) a feature kernel to the feature map in a stepwise manner, wherein the feature kernel is trained to identify features based on a feature map of a first feature map size, wherein the feature kernel has a first feature map size , where the feature map is obtained by convolving a convolution kernel on an image with the first image size, which makes, at least due to the convolution, the feature map with a second feature map size, where the stepping is determined by larger than the first feature map size half of the step size representation. A corresponding computer program (703) and computer program carrier (705) are also disclosed.

Description

technical field [0001] Embodiments herein relate to image processing by using convolutional neural networks. In particular, methods and image processing entities for applying a convolutional neural network to images of a second image size are disclosed. A corresponding computer program and computer program carrier are also disclosed. Background technique [0002] Within the field of image processing, a particular genre involves object detection via convolutional neural networks. The use of convolutional neural networks has become increasingly popular, for example due to their computational efficiency. [0003] Known convolutional neural networks configured to recognize objects (eg, cars, cats, people, etc.) in an image take that image as input and provide a score, such as the probability that the object is present in the image. [0004] In the so-called convolution action performed to analyze an image, a filter kernel is applied in a sliding window fashion over the entire...

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 Patents(China)
IPC IPC(8): G06K9/62G06N3/04G06V10/24
CPCG06N3/045G06F18/21G06V10/24G06V10/454G06T7/97G06T7/44G06N3/08
Inventor N·丹尼尔松S·莫林M·斯堪斯
Owner AXIS