A human activity recognition method based on sparse representation and Softmax classification

Abstract

The invention discloses a human activity identification method based on sparse representation and Softmax classification. The method comprises: Step S1: a Softmax model being trained by using a largeamount of wireless sensor activity signal data set. Step S2: classifying and identifying the real-time detected activity signals using the trained Softmax model. By adopting the technical proposal ofthe invention, the massive sensor activity signal data set in the database is trained, the belonging class of the signal is determined, and then the activity signal is processed by combining the sparse representation algorithm, thereby greatly reducing the calculation amount and the complexity of the identification of the human activity signal, and simultaneously effectively improving the accuracyof the human activity analysis.

Description

technical field [0001] The invention relates to the field of medical human body activity detection and recognition, in particular to a human body activity recognition method based on sparse representation and Softmax classification. Background technique [0002] In the past few decades, the advancement of modern computer technology has made human activity recognition a popular research field, and the analysis of human activity recognition has great research significance in the fields of medicine, security, and human-computer interaction. Most current activity recognition algorithms are based on activity recognition under wireless sensor networks. Wireless sensors collect activity signals, then transmit and process the collected activity signals, and then use signal processing algorithms for identification. However, due to the large amount of noise in the wireless sensor signal and the huge amount of storage and calculation of the signal, there are certain challenges to the r...

AI Technical Summary

Problems solved by technology

[0004] 1. The acquisition of activity signals is affected by external factors, and there are a large number of noise signals, which have a very important impact on the accuracy of subsequent algorithm processing

[0005] 2. The calculation of activity signal storage is huge, and the processing of recognition algorithms often requires a large time complexity, and there is a large time delay, which cannot effectively identify activities in real time

Images