Object Emotion Analysis Model Fusing Static and Dynamic Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing emotion analysis methods face challenges in accurately analyzing emotions due to interference factors such as illumination, occlusion, and personalized facial features, leading to low accuracy in emotion analysis results.
Innovation Solution
The proposed object emotion analysis method involves acquiring multimedia data, extracting static facial features and dynamic features (including expression change, sound, and language content features), and inputting these features into a pre-trained object emotion analysis model to fuse and output an emotion analysis result.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If facial features are extracted from images for emotion analysis, then emotion analysis can be performed, but interference factors such as illumination, occlusion, and personalized facial features reduce the accuracy of the analysis results
Solution Approach 1:
The patent extracts and separates emotion-related feature information from facial features while removing interference factors such as illumination, occlusion, and personalized facial features. This is achieved through a deep learning model that specifically isolates emotion-characterizing features from the complex facial data, thereby improving emotion analysis accuracy by eliminating harmful interference elements.
Solution Approach 2:
The patent transforms the approach by changing from analyzing raw facial features to analyzing extracted emotion-related feature information. The deep learning model performs parameter transformation by converting complex facial feature data into simplified emotion-specific parameters, which are less susceptible to interference from illumination, occlusion, and personalized features.
2Measurement precision
If deep learning models learn from facial features containing interference factors, then emotion analysis can be performed, but the interference factors influence the learning process and reduce result accuracy
Solution Approach 1:
The patent extracts pure emotion-related feature information from facial features by removing interference factors before feeding them to the deep learning model. This extraction process ensures that the model learns from clean, high-quality emotion-specific features rather than noisy raw facial data, improving both learning effectiveness and analysis accuracy.
Solution Approach 2:
The patent performs preliminary processing of facial features to extract emotion-related information before the main emotion analysis task. By pre-processing the data to isolate emotion-specific features and remove interference, the system prepares high-quality input for the deep learning model, thereby improving the overall learning process and final accuracy.
Data Source
AI summary
An object emotion analysis method and apparatus and an electronic device are provided. The method includes: extracting a static facial feature and a dynamic feature from multimedia data associated with a target object, wherein the dynamic feature includes one or more of an expression change feature, a sound feature and a language content feature: inputting the static facial feature and the dynamic feature into a pre-trained object emotion analysis model, fusing the static facial feature and the dynamic feature by the object emotion analysis model, and outputting an emotion analysis result.