An improved neural network algorithm for acceleration sensor to recognize human posture. (3rd April 2018)

• Record Type: Journal Article
• Authors: Lv, Liping
• Is Part Of: Journal of discrete mathematical sciences & cryptography. Volume 21:Number 3(2018)
• Page Start: 789
• Page End: 795
• Publication Date: 2018-04-03
• Subjects: Acceleration sensor -- Recognition -- Human posture -- Improved neural network; Computer science -- Mathematics -- Periodicals; Cryptography -- Periodicals; Computer science -- Mathematics; Cryptography; Periodicals; 004.0151
• DOI: 10.1080/09720529.2018.1449318
• Languages: English
• ISSNs: 0972-0529

Abstract

Abstract: The computational complexity of traditional human body recognition algorithm is higher, and the recognition effect is poor. Therefore, in this paper, an improved neural network algorithm for acceleration sensor too recognize human posture is proposed. First of all, the sliding window method is used for human posture segmentation; then, a neuron model for acceleration sensor to recognize human posture is established; finally, the pooling layer is introduced to improve the convolution neural network to prevent overfitting during recognition. Experimental results show that the proposed algorithm can improve the recognition rate and computational complexity, and has better recognition results.