Behavior Prediction Using Graph Data and Relationship Rules

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current behavior recognition techniques are limited in that they only recognize past or current behaviors, making it too late to take preventive measures, as they do not anticipate future behaviors or situations that require countermeasures.

Innovation Solution

An information processing program and apparatus that acquires video data, specifies relationships between objects, recognizes human behavior, and predicts future behaviors by comparing the behavior with pre-defined rules associated with relationships, allowing for proactive detection of situations needing countermeasures.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If behavior recognition technique is used to recognize current or past behavior, then behavior recognition accuracy is improved, but response time is too late to take preventive measures

Engineering Contradiction:
Improvebehavior recognition accuracyVSAvoidresponse time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary action by predicting future behavior before it actually occurs. The behavior prediction unit uses graph data representing relationships between objects and behavior prediction rules to forecast what a person will do next, enabling preventive measures to be taken in advance rather than reacting after the behavior occurs.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies beforehand cushioning by preparing prediction models and relationship data in advance. The graph data structure and behavior prediction rules are pre-established to cushion against potential harmful behaviors, allowing the system to anticipate and prepare for future situations rather than reacting too late.

Inventive Principle:
Principle #11Beforehand cushioning (Prior cushioning)

2Adaptability or versatility

If behavior prediction rule and graph data are integrated, then predictive capability is improved, but system complexity increases

Engineering Contradiction:
Improvepredictive capabilityVSAvoidsystem complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system applies segmentation by dividing the complex prediction task into distinct components: a graph data structure for representing relationships between objects, behavior recognition units for identifying current actions, and behavior prediction units for forecasting future behaviors. This modular segmentation makes the overall system more manageable and implementable despite its predictive capabilities.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4207107A1Information processing program, information processing method, and information processing apparatus
Publication Date: 2023.07.05 FUJITSU LTD
  • EP4207107A1 patent drawingFigure 1
  • EP4207107A1 patent drawingFigure 2
  • EP4207107A1 patent drawingFigure 3

AI summary

An information processing program that causes at least one computer to execute a process, the process includes acquiring (S101) video data that has target objects that include a person and an object; specifying (S103) each relationship of each of the target objects in the acquired video data by using graph data that indicates a relationship of each of target object stored in a storage unit; specifying (S106) a behavior of the person in the acquired video data by using a feature of the person; and predicting (S108) a future behavior of the person or a state of the person by comparing the specified behavior of the person and the specified relationship with a behavior prediction rule, the behavior prediction rule being a rule associated with a future behavior of a person for each combination of a behavior of the person and a relationship.