Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Image label determination method and device and terminal

A technology for image labeling and determining methods, which is applied in the field of image processing and can solve problems such as poor flexibility of image labeling

Active Publication Date: 2018-10-16
BEIJING DAJIA INTERNET INFORMATION TECH CO LTD +1
View PDF5 Cites 23 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] Embodiments of the present invention provide an image label determination method, device, and terminal to solve the problem of poor flexibility in image label prediction in the single-label classification method and multi-label classification method in the prior art

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
  • Image label determination method and device and terminal
  • Image label determination method and device and terminal
  • Image label determination method and device and terminal

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] refer to figure 1 , shows a flow chart of steps of a method for determining an image label according to Embodiment 1 of the present invention.

[0033] The image label determination method of the embodiment of the present invention may include the following steps:

[0034] Step 101: input the image into the convolutional neural network, and determine the feature map of the image.

[0035] The image in this embodiment of the present invention may be a single frame image in a video, or just a multimedia image. An image is input into a convolutional neural network, and a feature map is obtained after passing through a convolutional layer or a pooling layer.

[0036] For the specific processing method of inputting the image into the convolutional neural network to obtain the feature map, refer to the existing related technology, which is not specifically limited in the embodiment of the present invention.

[0037] Step 102: Input the feature map into the classifier.

[00...

Embodiment 2

[0048] refer to figure 2 , shows a flow chart of steps of a method for determining an image label according to Embodiment 2 of the present invention.

[0049] The image tag determination method in the embodiment of the present invention may specifically include the following steps:

[0050] Step 201: Construct sample images corresponding to each classification task.

[0051] When constructing sample images corresponding to each classification task, the sample images under each classification task are collected, and each sample image is labeled. Specifically, when labeling sample images, for each sample image, determine the classification task to which the sample image belongs, and the target label corresponding to the sample image under the classification task; set the probability value of the sample image at the target label position to is 1, and the probability value of the sample image at the label position except the target label under each classification task is set to...

Embodiment 3

[0083] refer to image 3 , shows a structural block diagram of an apparatus for determining an image label according to Embodiment 3 of the present invention.

[0084] The image label determination device in the embodiment of the present invention may include: a determination module 301 configured to input an image into a convolutional neural network to determine a feature map of the image; an input module 302 configured to input the feature map into a classification In the device; wherein, the classifier includes a plurality of classification tasks, and each classification task is a multi-label classification task or a single-label classification task; the prediction module 303 is configured to pass each classification task in the classifier, respectively Predict the label corresponding to the feature map; wherein, each label corresponds to a probability value; the label determination module 304 is configured to filter the labels of the image from the predicted labels accordi...

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 invention provide an image label determination method and device and a terminal. The method comprises the following steps of: inputting an image into a convolutional neural networkand determining a feature map of the image; inputting the feature map into a classifier, wherein the classifier comprises a plurality of classification tasks, and each classification task is a multi-label classification task or a single-label classification task; respectively predicting labels corresponding to the feature map through each classification task in the classifier, wherein each label corresponds to a probability value; and screening a label of the image from the predicted labels according to the probability values. The image label determination method provided by the invention is capable of adaptively adjusting image classification manners according to image types, is strong in flexibility and is capable of enhancing the correctness of label prediction results.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method, device and terminal for determining an image label. Background technique [0002] Deep learning has been widely used in video images, speech recognition, natural language processing and other related fields. As an important branch of deep learning, convolutional neural network has greatly improved the accuracy of prediction results obtained in computer vision tasks such as target detection and classification due to its strong fitting ability and end-to-end global optimization ability. Although convolutional neural network-based algorithms on standard datasets have surpassed human recognition ability, these algorithms can only give an image a label. [0003] However, in real application scenarios, the general image includes multiple objects, and users often expect to be able to give multiple classification labels to the input image. At this time, using ...

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): G06K9/62G06N3/04G06V10/764G06V10/774G06V20/00
CPCG06N3/045G06F18/24G06N3/08G06V20/00G06V10/454G06V10/764G06V10/7715G06V10/774G06F18/2413G06N3/04G06F18/214G06F18/2415
Inventor 张志伟杨帆
Owner BEIJING DAJIA INTERNET INFORMATION TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com