Behavior Estimation Using Person-Object Relevance Aggregation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing behavior estimation techniques fail to accurately estimate a person's behavior due to the lack of consideration for the relevance between the person and peripheral objects, leading to reduced estimation accuracy, especially when objects are not detected.
Innovation Solution
A behavior estimation device that extracts features of a person, objects, and their periphery, integrates and aggregates these features using neural networks, and estimates behavior based on the aggregated information, incorporating relevance degrees between objects and persons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the existing behavior estimation technique is used that does not consider relevance between person and peripheral object, then the device complexity is reduced, but the estimation accuracy deteriorates
Solution Approach 1:
The behavior estimation device is segmented into multiple independent functional units: a person feature extraction unit, an object feature extraction unit, a peripheral feature extraction unit, a feature aggregation unit, and a behavior estimation processing unit. Each unit processes specific features independently before aggregation, which improves estimation accuracy by considering relevance between persons and objects while maintaining manageable complexity through modular design.
2Measurement precision
If the feature aggregation processing is executed to aggregate person feature, object feature, and peripheral feature, then the estimation accuracy is improved, but the processing time increases
Solution Approach 1:
The device performs preliminary feature extraction independently for persons, objects, and periphery before aggregation. By preparing these features in advance through dedicated extraction units, the system reduces the complexity and time required for the final aggregation and behavior estimation processing, thus improving overall processing efficiency while maintaining high estimation accuracy.
3Measurement precision
If the relevance between person and object is considered in behavior estimation, then the estimation accuracy is improved, but the difficulty of detecting and measuring increases
Solution Approach 1:
The feature aggregation unit acts as an intermediary that receives independently extracted features from person, object, and peripheral detection units. It processes these features to determine relevance relationships without requiring direct complex interaction between detection units, thereby improving estimation accuracy while managing the difficulty of detecting and measuring relevance through a dedicated intermediate processing layer.
Data Source
AI summary
In the behavior estimation device, a person feature extraction means extracts a feature of a person detected from a plurality of images in a time series. An object feature extraction means extracts a feature of an object detected from the plurality of images. A peripheral feature extraction means extracts a feature of a periphery of the person in the plurality of images. A feature aggregation means executes aggregation processing for aggregating the feature of the person, the feature of the object, and the feature of the periphery of the person. A behavior estimation processing means executes estimation processing for estimating the person's behavior included in the plurality of images based on information including a processing result of the aggregation processing.