Posture Guide Using Machine Learning on 2D Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing posture analysis services rely on expensive devices like 3D cameras or sensors, and provide non-customized feedback that cannot be dynamically adjusted to a user's current state, lacking real-time guidance and user-specific feedback.
Innovation Solution
A method using a machine learning model to analyze a user's posture from a 2D camera image, generating personalized tutoring images and guide information that adjusts in real-time based on the user's posture, without the need for specialized hardware.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive devices such as 3D cameras or sensors are used to analyze user posture, then measurement precision and reliability are improved, but device complexity and cost increase
Solution Approach 1:
The patent uses a 2D camera to capture images and creates a digital copy or representation of the user's posture through image processing and machine learning algorithms. Instead of using complex 3D cameras or sensors, the system processes 2D images to extract posture information, effectively replacing expensive hardware with software-based solutions that analyze visual data.
Solution Approach 2:
The patent replaces mechanical sensing systems (3D cameras, sensors) with an optical system (2D camera) combined with computational methods. The machine learning model processes 2D images to infer 3D posture information, substituting physical sensing mechanisms with algorithmic analysis of visual data.
2Device complexity
If simple 2D camera images are used without machine learning, then device complexity is reduced, but the ability to provide customized and dynamic posture feedback is lost
Solution Approach 1:
The system uses machine learning models that automatically analyze 2D camera images to extract posture information without requiring manual intervention or complex device setup. The algorithm autonomously processes the visual data, identifies body joints and posture, and generates feedback, enabling the simple 2D camera system to provide adaptive, customized posture guidance.
3Productivity
If reference posture is simply compared with user posture, then processing speed is improved, but feedback customization and dynamic adjustment to user's current state are reduced
Solution Approach 1:
The patent implements a feedback mechanism where the machine learning model continuously analyzes the user's current posture from 2D images and compares it with reference postures to generate customized feedback. The system dynamically adjusts the feedback based on the user's real-time state, providing actionable guidance that adapts to the user's progress and specific posture deviations.
Data Source
AI summary
According to an embodiment of the present invention, there is provided a posture guide provision method performed by an apparatus for providing posture guide using a preset machine learning model. The provision method comprises acquiring an image for a user's posture, displaying a tutoring image in a first region of a display and the acquired image in a second region of the display, extracting a feature point from the acquired image, acquiring user posture information by generating a user posture line corresponding the user's posture based on the extracted feature point, generating posture guide information for guiding the user's posture based on the tutoring image and the user posture information, and combining the acquired image and the posture guide information with each other and displaying the combination in the second region of the display.