AI Dance Matching System for Movement Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional dance training devices struggle to effectively provide users with a variety of songs and accurately evaluate dance movements, particularly failing to capture upper body movements and stimulate competitive spirit through limited song offerings.
Innovation Solution
A dance matching method and system that extracts a group dance part from a reference song video, calculates a matching rate between user and reference dance videos, and transmits this rate to the user's device, allowing for real-time editing and provision of various songs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional dance training devices use arrow platforms to guide dance movements, then users can follow dance movements through visual cues, but the devices cannot capture upper body movements and are limited in providing diverse song evaluations
Solution Approach 1:
The patent replaces the mechanical arrow platform system with an AI-based video analysis system. Instead of using physical arrows on a platform to guide movements, the system uses computer vision and deep learning algorithms to automatically recognize and evaluate dance movements from video footage, enabling comprehensive tracking of both lower and upper body movements across diverse songs
Solution Approach 2:
The patent creates a multi-functional system that can handle various song types, dance styles, and movement complexities through a single AI-based evaluation platform. The system universally processes different dance genres and song formats, providing accurate movement recognition and scoring without requiring separate specialized devices for each song type
2Adaptability or versatility
If dance training devices provide limited song selections, then the device complexity remains manageable, but the user experience lacks variety and competitive stimulation
Solution Approach 1:
The patent implements a system where the AI automatically extracts dance movements, generates evaluation criteria, and processes video data without requiring manual configuration for each song. The system self-adapts to new songs and dance styles, automatically learning movement patterns and providing accurate evaluations, thereby providing unlimited song variety without proportionally increasing operational complexity
Solution Approach 2:
The patent uses parameter-based AI models that can dynamically adjust evaluation criteria based on different song characteristics and dance styles. By changing computational parameters rather than physical device configuration, the system accommodates diverse songs and dance types without increasing hardware or system structural complexity
Data Source
AI summary
A dance matching method, according to one embodiment of the present invention, which relates to a method for evaluating a dance of a user on the basis of a reference song video having a dance of multiple dancers, may comprises the steps in which: a group dance part extraction unit extracts a group dance part from the reference song video under a preset dance extraction condition; a song extraction unit extracts a song part corresponding to the extracted group dance part; a communication unit transmits the song part to an electronic device of the user so that the extracted song part is played back in the electronic device; the communication unit receives a user dance video from the electronic device while the song part is being played back in the electronic device; a matching unit calculates a matching rate between a dance of the user in the user dance video and a dance of the dancers in the reference song video; and the communication unit transmits the matching rate to the electronic device.