Exercise Motion Inference Using Segmented Time-Series Elements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing exercise assisting apparatuses face challenges in easily generating form models and inference models due to the need for large quantities of training data for new exercises not previously learned, making it difficult to adapt to diverse motion patterns.
Innovation Solution
The apparatus divides motion information into motion elements based on regularity and uses a pre-trained model to learn time-series antecedent dependencies, enabling the generation of inference models that can infer motions from target motion information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a new form model is generated through machine learning for exercises not yet learned, then the adaptability to diverse motion patterns is improved, but the difficulty of model generation increases due to the need for large quantities of training data
Solution Approach 1:
The patent divides motion information into multiple motion elements based on regularity, allowing the model to learn and generalize from segmented motion patterns. This segmentation enables the system to handle diverse exercises without requiring complete training data for each exercise type, as the segmented motion elements can be recombined to infer new exercise patterns.
Solution Approach 2:
The patent uses a pre-trained model that has already learned motion patterns from existing training data. This preliminary training allows the system to infer motions for new exercises by applying the pre-learned motion elements and temporal relationships, eliminating the need to train from scratch for each new exercise type and significantly reducing the data quantity required.
2Ease of manufacture
If a pre-trained model is used to learn time-series antecedent dependency, then the ease of model generation is improved, but the complexity of the learning process increases due to time-series analysis requirements
Solution Approach 1:
By segmenting motion information into discrete motion elements with temporal relationships, the patent transforms complex continuous motion data into manageable units. This segmentation simplifies the learning process by allowing the model to focus on learning temporal dependencies between discrete elements rather than processing entire complex motion sequences at once.
Solution Approach 2:
The pre-trained model performs preliminary learning of motion patterns and temporal relationships during an initial training phase. This preliminary action encapsulates the complexity of time-series analysis in the pre-trained model, allowing subsequent inference for new exercises to be performed more simply by applying the pre-learned patterns to new motion data.
Data Source
AI summary
An exercise assisting apparatus 1 includes: a division unit 312 configured to divide sample motion information 3231 indicating motions of a sample person doing exercise into a plurality of pieces of motion element information 3235, according to regularity of the motions of the sample person; and a generation unit 314 configured to generate an inference model by causing a pre-trained model 322 to learn time-series antecedent dependency of the plurality of pieces of motion element information, the inference model inferring motions of a target person, based on target motion information 3211 indicating the motions of the target person doing exercise.