In-Vehicle Pose Detection for Accurate Seat Assignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing seat occupancy detection systems in vehicles face challenges in accurately assigning a person to a seat due to unclear positions or movements, leading to false assignments, which can impact safety controls such as airbag regulation and seatbelt tensioning.
Innovation Solution
A method and system for pose detection of a person in a vehicle using on-board cameras, which involves determining face and body bounding areas, calculating characteristics, and updating seat assignment logic based on the person's leaning position to improve seat occupancy classification and safety control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional object detection is used to assign persons to seats based on fixed area positions, then the system is simple to implement, but wrong seat assignments occur when persons are in unclear positions or move
Solution Approach 1:
The patent transitions from 2D object detection to 3D pose detection by introducing depth information and spatial relationships. Instead of only analyzing flat images, the system uses multiple camera views and geometric calculations to determine the three-dimensional position and orientation of persons, enabling accurate seat assignment even when persons lean or move within the vehicle
Solution Approach 2:
The system dynamically adjusts seat assignments based on real-time pose detection. Rather than using fixed detection zones, the system continuously monitors person position, leaning angle, and body orientation, adapting the assignment logic to match the current dynamic state of the person in the vehicle
2Reliability
If only visible part of person is considered due to occlusion, then the detection process is simplified, but false seat assignments result when person covers part of their body
Solution Approach 1:
The patent segments the person's body into multiple detectable parts (head, torso, limbs) and uses the spatial relationships between these segments to infer complete pose information. Even when parts are occluded, the detection system can reconstruct the full body position and orientation by analyzing the visible segments and their geometric relationships
Solution Approach 2:
The system introduces intermediate geometric models and pose estimation algorithms as mediators between the captured images and the final seat assignment. These intermediaries fill in missing information from occluded body parts by using constraints from visible parts and known human body geometry, enabling accurate detection despite occlusions
Data Source
Figure 1
Figure 2A~2B
Figure 3~4
AI summary
A computerized method of pose detection of a person in a vehicle is presented. The method comprises receiving, from an on-board camera, an image of an interior of the vehicle showing a seat of the vehicle occupied by the person, obtaining at least one first characteristic of a first face bounding area and a first body bounding area associated with the occupied seat of the vehicle, determining a second body bounding area and an associated second face bounding area of the person from the image, determining at least one second characteristic of the second face bounding area and the second body bounding area, and determining a pose of the person based on at least one second characteristic and on at least one first characteristic.