Vehicle Occupant Counting via Multi-View Coordinate Grouping

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting the number of persons on a vehicle are inaccurate due to variations in seating layouts and reliance on pre-held information about seating arrangements, which can lead to insufficient accuracy in counting occupants.

Innovation Solution

An analysis apparatus that continuously images a vehicle from different directions, detects predetermined vehicle parts and human faces, assigns coordinates in a vehicle coordinate system, groups faces with similar coordinates, and counts these groups to determine the number of persons, eliminating the need to identify specific seats.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If pre-held seating layout information is used to detect persons, then detection process is simplified, but detection accuracy deteriorates due to variations in actual seating arrangements

Engineering Contradiction:
Improvedetection process simplicityVSAvoidperson counting accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The invention extracts only the necessary information (coordinates of detected persons) from multiple images without requiring pre-held seating layout information. By taking out the dependency on predetermined seating arrangements, the system achieves both simplified operation and improved accuracy through coordinate-based grouping of detected persons across multiple images.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If multiple images are continuously captured to improve detection accuracy, then person counting precision improves, but processing time and system complexity increase

Engineering Contradiction:
Improveperson counting accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary detection of person coordinates in multiple continuously captured images before final counting. By detecting coordinates in advance across multiple images and then grouping them, the system prepares data structure beforehand, which streamlines the final counting process and reduces overall processing time while maintaining high accuracy.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The invention uses coordinate information as a simplified copy or representation of person positions instead of processing entire images for counting. By working with coordinate data rather than full image processing, the system significantly reduces computational complexity and processing time while preserving detection accuracy through coordinate-based grouping.

Inventive Principle:
Principle #26Copying

3Measurement precision

If coordinate-based grouping method is used instead of seat identification, then detection accuracy improves by avoiding seating layout assumptions, but coordinate system establishment complexity increases

Engineering Contradiction:
Improvedetection accuracyVSAvoidcoordinate system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The coordinate system established in the invention serves multiple functions: it provides a reference framework for detecting person positions, enables grouping of detected persons across different images, and eliminates dependency on specific seating layout information. This universal coordinate-based approach works for various vehicle types and seating arrangements, achieving high detection accuracy without requiring complex seat identification mechanisms.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3651114B1Analysis device, analysis method, and program
Publication Date: 2022.06.08 NEC CORP
  • EP3651114B1 patent drawingFigure 1
  • EP3651114B1 patent drawingFigure 2
  • EP3651114B1 patent drawingFigure 3

AI summary

According to the present invention, there is provided an analysis apparatus (10) including an image analysis unit (11) that detects a predetermined part of a vehicle and persons on the vehicle from each of a plurality of images obtained by imaging the same vehicle from different directions a plurality of times, and detects coordinates of each of the plurality of persons in a coordinate system having the detected predetermined part as a reference; a grouping unit (12) that groups the persons detected from the different images, on the basis of the coordinates; and a counting unit (13) that counts the number of groups.