Hybrid Posture Detection Framework: Connecting Deep Neural Networks and Machine Learning
N. Jeevana Jyothi, K. Lakshmi Devi, T. Praneetha, R. Annapurna, V. Siva, S. Naveen · Advances in computer science research · 2024
Many researchers in the fields of artificial intelligence and human sensing have been attempting to find a solution to the issue of posture detection.Posture recognition for the purpose of remote geriatric health monitoring, including standing, sitting, and walking.Most recent research has used conventional ML classifiers for posture recognition.When these algorithms are used for posture detection, the accuracy drops a bit.An innovative hybrid method for posture detection has been created by combining ML classifiers such as Support Vector Machine (SVM), Logistic Regression (KNN), Decision Tree, Naive Bayes, Random Forest, Linear Discrete Analysis, and Quadratic Discrete Analysis with DL classifiers such as Long Short-Term Memory (LSTM) and Bidirectional LSTM, 2Dconvolutional Neural Networks (2D-CNN), and 1D-ConvolutionalNetworks (1DNN).The goal of combining DL and ML algorithms in a hybrid fashion is to boost their prediction capabilities.With our experiments on a popular benchmark dataset, we achieved an accuracy of 98% or better.