Computer Vision Motion Analysis for Depth-Aware Exercise Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing motion tracking systems using optical devices struggle to accurately determine the positioning and movement of body parts that extend predominantly in the depth dimension, leading to errors and inaccuracies in computer vision techniques.

Innovation Solution

A method that utilizes calibration images to establish reference length values for body members, computes length variation factors, and applies trigonometric relationships to determine scaled length values, enabling accurate estimation of depth-related parameters through optical device processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If computer vision techniques process images to determine body member orientations and motion, then motion tracking capability is enabled, but measurement precision deteriorates when body members extend in the depth dimension perpendicular to the image plane

Engineering Contradiction:
Improvemotion tracking capabilityVSAvoiddepth dimension measurement accuracy
Core Design Contradiction:
Extent of automationVSMeasurement precision

Solution Approach 1:

The patent applies trigonometric relationships to calculate depth dimension parameters from 2D image measurements. By using calibration data and applying sine/cosine functions to the measured lengths and angles, the system reconstructs 3D depth information from planar images, effectively adding the depth dimension back into the analysis.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

Solution Approach 2:

The patent introduces calibration data as an intermediary element. By capturing known reference objects or body configurations at calibrated positions, the system creates a reference framework that mediates between 2D image measurements and 3D depth reconstruction, enabling accurate depth estimation through comparison with calibrated reference values.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If stereo cameras or depth sensors are used to improve depth dimension detection, then measurement precision improves, but device complexity and cost increase

Engineering Contradiction:
Improvedepth dimension measurement accuracyVSAvoidcamera system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent creates a virtual 3D model by calculating depth parameters from 2D image data through trigonometric relationships. Instead of using physical depth sensors, the system generates a computational copy of the 3D depth information that can be derived from standard 2D camera images combined with calibration data and mathematical transformations.

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces the mechanical/optical depth sensing approach (stereo cameras, depth sensors) with a computational/mathematical approach. By substituting physical depth measurement hardware with trigonometric calculations based on 2D image measurements and calibration data, the system achieves depth reconstruction without complex sensing hardware.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If calibration images are processed to establish reference length values, then measurement precision improves for depth dimension, but loss of time increases due to additional processing steps

Engineering Contradiction:
Improvedepth dimension measurement accuracyVSAvoidcalibration and processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs calibration data capture and processing in advance, before actual motion tracking begins. By establishing reference length values and calibration parameters upfront, the system creates a reusable framework that speeds up subsequent depth calculations during exercise monitoring, as the computationally intensive calibration step only needs to be done once rather than continuously.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4383194B1Computer vision-assisted analysis of motion of a person
Publication Date: 2026.04.15 SWORD HEALTH SA
  • EP4383194B1 patent drawingFigure 1A~1B
  • EP4383194B1 patent drawingFigure 2~3
  • EP4383194B1 patent drawingFigure 4~5B

AI summary

A method comprising: digitally processing at least one calibration image and at least one exercising image of a person taken with an optical device to compute at least one distance between predetermined anatomical landmarks of the person so that at least one calibration length value and at least one target length value per body member of at least first and second predetermined body members of the person are computed; digitally computing a length variation factor based on both the at least one calibration length value of the first predetermined body member and the at least one target length value of the first predetermined body member; digitally computing a scaled length value associated with the second predetermined body member; and digitally computing at least one parameter of the second predetermined body member based on both the scaled length value and the at least one target length value of the second predetermined body member, the at least one parameter comprising: a difference in depth of the at least one predetermined anatomical landmark relative to at least one other predetermined anatomical landmark, and/or a range of motion in depth of the at least one predetermined anatomical landmark relative to the at least one other predetermined anatomical landmark.