A Complex Activity Recognition Method Fused with Motion, Physiology, and Position Sensing Data

Abstract

The invention discloses a complex activity recognition method that integrates motion, physiology and position sensing data, including: 1) collecting motion, physiology and position sensing data; 2) extracting statistical features from the motion sensing data, and performing a K-Means clustering and LDA topic extraction to obtain the topic distribution of motion sensing data, and establish a sub-classifier based on motion sensing data; 3) extract structural features and transient features from physiological sensing data, and establish a physiological Sub-classifier for sensing data; 4) Divide the city map into grid units, introduce Weibo check-in, ATUS and POI data sets, use matrix decomposition to establish the relationship between grid units and complex activities, and use user location sensing data Mapping to the corresponding grid unit to obtain sub-classifiers based on position sensing data; 5) The outputs of these sub-classifiers are fused using a fractional fusion method to obtain the probability of complex activities.

Description

technical field [0001] The invention relates to the field of activity recognition, in particular to a complex activity recognition method that integrates motion, physiology and position sensing data. Background technique [0002] With the development of smart devices (such as smartphones, smart watches, etc.) , and the acquisition of location-related sensor data such as GPS points has become increasingly convenient. How to use these data for activity recognition has become the focus of the industry. The corresponding technology has broad application prospects in the fields of healthcare, elderly assistance, and skill assessment. . [0003] Human activities can be divided into simple and complex activities. Simple activities usually consist of periodic movements or single postures of the human body, such as standing, sitting, walking, running, etc. Compared with simple activities, complex activities are usually less regular, last longer, and have high-level semantics, such...

AI Technical Summary

Problems solved by technology

[0008] However, these methods either require a large amount of labeled data, or do not make full use of user place information, and the activity recognition model cannot work when a user visits a new type of place

Images