Image sentiment analysis method based on joint attribute modeling

A sentiment analysis and attribute modeling technology, applied in neural learning methods, character and pattern recognition, biological neural network models, etc., can solve the problem of ignoring the correlation between middle-level semantic concepts and emotional semantics, unable to effectively solve the semantic gap, ignoring semantics information and other issues, to enhance the ability of semantic representation and discrimination, reduce the semantic gap, and improve the accuracy and robustness.

Pending Publication Date: 2022-03-18
SHANGHAI UNIV
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the existing methods based on middle-level semantic features widely cover the concrete content of images, many concepts that contribute to emotional transmission are missing in the semantic space covered by general semantic concepts, and only rely on the semantic concepts covered by semantic concept sets. Visual content, ignoring some semantic information that cannot be expressed in language, has certain limitations and cannot effectively solve the problem of semantic gap
In addition, the deep learning method driven by big data is also the main technical idea of ​​current research, but the extracted features with emotional semantics require large-scale and low-noise data for training, and ignore the correlation between middle-level semantic concepts and emotional semantics. sex, lack of interpretability

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 sentiment analysis method based on joint attribute modeling
  • Image sentiment analysis method based on joint attribute modeling
  • Image sentiment analysis method based on joint attribute modeling

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0067] An image sentiment analysis method based on joint attribute modeling, which uses emotional attribute labels as supervisory information to obtain the corresponding local visual regions to learn emotional attribute features; based on Laplacian feature mapping to mine inter-class information, obtain discriminative The emotional potential attribute features; and add an optimization algorithm to the dictionary learning framework, and finally use the learned joint emotional attribute features as middle-level features for final image sentiment analysis. The operation steps are as follows:

[0068] Step 1: Mining emotional attributes based on social media user metadata information, on this basis, constructing emotional attribute sets by studying the relationship between semantic concepts and emotions, and then obtaining image vision through a model that integrates neural networks and matrix decomposition The correlation between features and emotional attributes completes the emo...

Embodiment 2

[0076] This embodiment is basically the same as the above-mentioned embodiment, and the special features are:

[0077] In this embodiment, a flow chart of an image sentiment analysis method based on joint attribute modeling, see figure 1 , the method includes the following steps:

[0078] Step 1: Emotional attribute mining based on user metadata information and image label prediction of fusion neural network and matrix factorization specifically include the following steps:

[0079] Mining emotional attributes from emotional image data with user tags, constructing an emotional attribute set composed of semantic concepts in line with human emotional cognition; in image data sets with user tags, through a model that fuses neural networks and matrix decomposition The mapping from image visual features to emotional attribute labels is completed, and the emotional attribute labels of images are obtained.

[0080] Step 2: Construct different dictionaries to obtain emotional attrib...

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 image sentiment analysis method based on joint attribute modeling. A dictionary learning method is utilized to construct a sentiment attribute dictionary and a sentiment potential attribute dictionary to obtain joint sentiment attribute features. In order to ensure the effectiveness of the combined emotion attribute features obtained by dictionary learning on image emotion analysis, an emotion attribute tag is used as supervision information to obtain a local visual area corresponding to the emotion attribute tag, and the emotion attribute features are learned by applying constraint conditions; and on the basis of Laplacian feature mapping, obtaining discriminative emotion potential attribute features by mining inter-class relationships. According to the attribute modeling mode, the emotional property of the semantic concept is fully utilized, the semantic characterization capability of the joint emotion attribute characteristics is enhanced, and the discrimination capability of the joint emotion attribute characteristics is also ensured. And finally, providing a corresponding optimization algorithm for a dictionary learning framework of joint attribute modeling, and performing image emotion recognition by taking joint emotion attribute features obtained by learning as middle-level features.

Description

technical field [0001] The invention relates to the field of image emotion analysis, in particular to an image emotion analysis method based on joint attribute modeling. Background technique [0002] With the prosperity of social media and the popularization of mobile terminals with camera functions, image data are continuously pouring into the network, becoming one of the main media for users to express their emotions. As a supplement to text sentiment analysis, image sentiment analysis has also become a research hotspot in the field of image semantic understanding. Image sentiment analysis is a kind of abstract semantic understanding research on the perceptual layer and cognitive layer. Nowadays, the excellent performance of deep learning and other technologies in the field of computer vision has brought opportunities for the research of image emotion analysis. At the same time, the related applications of image emotion analysis The expansion also provides room for furthe...

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): G06F16/35G06K9/62G06N3/04G06N3/08G06V10/764G06V10/774G06V10/82
CPCG06F16/35G06N3/08G06N3/045G06F18/2411G06F18/214
Inventor 朱永华高文靖朱蕴文顾庭彦
Owner SHANGHAI UNIV
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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap