Vehicle Occupant Head Pose Detection Using Image Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional techniques for estimating the orientation of a vehicle occupant's head pose struggle with accuracy, especially when the head pose shifts abruptly outside the detection range, leading to unstable calculations of angular velocity.

Innovation Solution

A detection device and learning device that segment images of a vehicle interior into first and second regions, using feature amounts from these regions to accurately detect the orientation of a body part, particularly employing a learning model that correlates facial landmarks with the head pose and changes in the vehicle interior environment, such as seat exposure, to stabilize orientation detection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional head pose estimation techniques are used, then detection can be performed within a predetermined angle range, but accuracy deteriorates when head pose shifts abruptly outside the detection range

Engineering Contradiction:

Improvehead pose detection accuracyVSAvoiddetection stability during turn-back action

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The image is divided into multiple regions: a first region containing the face and a second region containing other body parts or vehicle interior. By segmenting the image and analyzing multiple regions separately, the system can detect head pose more reliably even during abrupt movements, as the second region provides additional reference information that stabilizes detection when the face alone is insufficient.

Inventive Principle:

Principle #1Segmentation

2Adaptability or versatility

If angular velocity calculation is used to estimate head pose outside detection range, then detection can be extended beyond predetermined angles, but calculation stability deteriorates during abrupt head pose changes

Engineering Contradiction:

Improvedetection rangeVSAvoidcalculation stability

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system introduces a second region (containing other body parts or vehicle interior features) as an intermediary reference that mediates between the face region and the head pose estimation. This intermediary provides stable reference points that help maintain calculation stability even when the head pose changes abruptly, replacing the need for unstable angular velocity calculations during turn-back actions.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Device complexity

If only facial region images are used for detection, then the detection process is simple, but detection accuracy deteriorates during turn-back actions

Engineering Contradiction:

Improvedetection process complexityVSAvoidorientation detection accuracy

Core Design Contradiction:

Device complexityVSMeasurement precision

Solution Approach 1:

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS10706301B2Detection device, learning device, detection method, learning method, and recording medium

Publication Date: 2020.07.07 PANASONIC INTELLECTUAL PROPERTY MANAGEMENT CO LTD

AI summary

A detection device includes an image segmenter and a detector. The image segmenter cuts out a first region image and a second region image from an image of a vehicle interior that is acquired from an imaging device. The first region image shows at least a portion of a first part of a body of an occupant. The second region image shows at least a portion of a region of the vehicle interior around the first part or at least a portion of a second part of the body of the occupant. The detector detects an orientation of the first part of the body of the occupant based on a feature amount of the first region image and a feature amount of the second region image.