A method for training action recognition

Abstract

A training action identification method comprises the steps of: measuring motion parameters of motion, wherein the motion parameters comprise an accelerated speed, an angular speed and a rotation attitude angle; according to a preset constraints model, performing denoising processing for the motion parameters to obtain optimal motion parameters; and employing a classification algorithm to performclassification for the optimal motion parameters to identify the action. According to the limb length and the limb connection relation, the constraints model is established, and the constraints modelis employed to perform denoising for the actually measured motion parameters to obtain the optimal motion parameters so as to allow the motion identification robustness to be higher.

Description

technical field [0001] The present application relates to the technical field of motion recognition, in particular to a training motion recognition method. Background technique [0002] In the process of physical training, it is necessary to scientifically evaluate the motion posture and frequency according to motion parameters (such as motion angular velocity, speed, cycle) to standardize training actions. [0003] "CN201610423325.X An upper limb movement recognition method for intelligent dumbbells" uses an inertial sensor module to collect and process the raw data of its acceleration sensor and gyroscope sensor, and then compare and analyze the data in the database corresponding to specific actions. Realize the recognition of dumbbell movement. [0004] Patent CN201610423325.X directly uses the raw data of the inertial sensor to compare with the standard action data in the database, which has the following disadvantages: 1. During the process of collecting standard actio...

AI Technical Summary

Problems solved by technology

[0004] Patent CN201610423325.X directly uses the raw data of the inertial sensor to compare with the standard action data in the database, which has the following disadvantages: 1. During the process of collecting standard actions, due to the noise of the sensor itself, the corresponding inertial raw data in the standard database may not be accurate enough; 2. Due to interference factors such as temperature, installation position error, and environment, the inertial sensor outputs raw data with a lot of noise in the actual exercise; 3. Theoretically, a single inertial sensor is more likely to produce standard movements similar to dumbbells in non-dumbbell training movements Corresponding sensor raw data; the above shortcomings will lead to misrecognition or missed recognition of dumbbell movement

Images