Supercharge Your Innovation With Domain-Expert AI Agents!

Optimization method of deep learning model

A technology of deep learning and optimization method, applied in the field of image processing, can solve the problem of not giving guidance relationship and so on

Inactive Publication Date: 2019-12-10
BEIJING SAMSUNG TELECOM R&D CENT +1
View PDF0 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this combination is only a combination at the feature level, and does not give a direct guiding relationship

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
  • Optimization method of deep learning model
  • Optimization method of deep learning model
  • Optimization method of deep learning model

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0026] The present invention proposes a self-supervised classification-assisted image segmentation method, which effectively combines global classification information and local segmentation by introducing a classification branch into the image segmentation network and using the classification output confidence to adjust the segmentation loss information. The image segmentation network consists of 3 modules: (1) shared base segmentation network, (2) unique classification branch and segmentation branch structure, (3) loss functions, including classification loss, segmentation loss and joint loss. Joint loss is a loss function defined jointly in terms of classification output and segmentation output.

[0027] figure 1 A segmentation network 100 using a deep learning model according to an embodiment of the invention is shown. The network structure 100 is composed of three parts, including a basic segmentation network 110 as the first part, two subtask branches 120 as the second...

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 provides an optimization method for a deep learning model, and the method comprises the steps: obtaining joint loss according to segmentation features and classification features; and optimizing the deep learning model according to the joint loss. According to the method, the global classification information can be effectively utilized to adjust the segmentation result of the localsegmentation region, and the problem that similar categories are easy to confuse is effectively avoided.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method for image segmentation and a computer-readable storage medium. Background technique [0002] Computer vision is widely used in the field of artificial intelligence (AI), which can be roughly divided into three directions: image classification (image-level), target detection (region-level) and semantic segmentation (pixel-level). In comparison, semantic segmentation is the most challenging, because it needs to classify each pixel, and if the category of each pixel is known, the image and the label of the region can be easily obtained. The application of semantic segmentation is very extensive, and it is the basic technology of many AI applications, such as unmanned driving, virtual fitting, intelligent robots, etc. As the mainstream technology of semantic segmentation, deep learning has achieved great improvement in both performance and speed. However, c...

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
IPC IPC(8): G06T7/11
CPCG06T7/11G06T2207/20081
Inventor 刘颖璐李春阳刘子坤熊君君
Owner BEIJING SAMSUNG TELECOM R&D CENT
Features
  • R&D
  • Intellectual Property
  • Life Sciences
  • Materials
  • Tech Scout
Why Patsnap Eureka
  • Unparalleled Data Quality
  • Higher Quality Content
  • 60% Fewer Hallucinations
Social media
Patsnap Eureka Blog
Learn More

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

© 2025 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com