Video Behavior Prediction Using Graph-Based Context Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing behavior recognition technologies are limited to recognizing behaviors that have already occurred, making it difficult to take timely countermeasures, such as preventing shoplifting or detecting early signs of illness, as they only react after the behavior has been performed.

Innovation Solution

An information processing system that analyzes video image data to predict future behaviors by combining behavior analysis with context sensing, using graph data to specify relationships between objects and individuals, and employing machine learning models like hidden Markov models to forecast future actions or states based on current behaviors and facial expressions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If behavior recognition technology is used to recognize current or past behaviors, then behavior detection capability is improved, but timeliness of countermeasure is worsened

Engineering Contradiction:
Improvebehavior detection capabilityVSAvoidtimeliness of countermeasure
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary action by predicting future behaviors before they occur. The behavior prediction unit uses the probability model to forecast future actions based on current behavior and contextual relationships, enabling countermeasures to be prepared in advance rather than reacting after the behavior has already happened.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies beforehand cushioning by preparing countermeasures in advance based on predicted future behaviors. The risk assessment unit evaluates potential risks from predicted behaviors and triggers appropriate countermeasures before the actual harmful behavior occurs, cushioning against potential negative outcomes.

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

2Device complexity

If simple behavior recognition is used, then system complexity is reduced, but prediction accuracy of future behavior is worsened

Engineering Contradiction:
Improvesystem complexityVSAvoidprediction accuracy of future behavior
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The system applies segmentation by dividing the complex behavior prediction task into distinct functional units: a behavior prediction unit that forecasts future actions, a relationship specification unit that identifies contextual relationships using graph data, and a risk assessment unit that evaluates potential risks. This modular segmentation manages system complexity while maintaining high prediction accuracy through specialized processing in each unit.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system introduces intermediary elements including graph data that represents contextual relationships between objects and individuals, and a probability model that serves as an intermediary between observed current behaviors and predicted future behaviors. These intermediaries enable accurate prediction without requiring direct complex processing of all possible behavior patterns.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Loss of time

If behavior prediction using probability models is implemented, then timeliness of countermeasure is improved, but device complexity is worsened

Engineering Contradiction:
Improvetimeliness of countermeasureVSAvoiddevice complexity
Core Design Contradiction:
Loss of timeVSDevice complexity

Solution Approach 1:

The system implements preliminary action by pre-establishing the probability model and graph data structures that enable rapid behavior prediction. Once these predictive mechanisms are in place, the system can quickly forecast future behaviors and trigger countermeasures without complex real-time analysis, improving timeliness while managing device complexity through pre-computed models.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4207101B1Information processing program, information processing method, and information processing apparatus
Publication Date: 2026.02.18 FUJITSU LTD
  • EP4207101B1 patent drawingFigure 1
  • EP4207101B1 patent drawingFigure 2
  • EP4207101B1 patent drawingFigure 3

AI summary

An information processing apparatus acquires (51) video image data that includes target objects including a person and an object, first specifies (52), by using graph data that indicates a relationship between each of target objects stored in a storage unit, a relationship between each of the target objects included in the acquired video image data, second specifies (53), by using a feature value of the person included in the acquired video image data, a behavior of the person included in the video image data, and predicts (54), by inputting the specified behavior of the person and the specified relationship to a probability model, a future behavior of a future state of the person.