Human posture recognition and analysis have been a widely studied topic these days because of wearable devices' innovation. Thus, activity tracking becomes an exciting use-case for healthcare and fitness tracking applications for both the elderly and adults. In this study, we present the analysis of several machine learning models to detect a human's posture by the data gathered by various accelerometers attached to the body. The following techniques/methods have been evaluated in this study:
- Support Vector Machines
- Logistic Regression
- Random Forest Classifier
- Stochastic Gradient Descent
- K-means clustering
- Gaussian Discriminant Analysis
- Neural Net (Multilayer Perceptron)
The dataset for this study is publicly made available by researchers at the Pontifical Catholic University of Rio De Janeiro [http://groupware.les.inf.puc-rio.br/har#dataset]. The dataset contains the following features:
- Name of the subject
- Gender of the subject
- Height of subject
- Body Mass Index of the subject
- X, Y, and Z-axis readings from 4 different accelerometers.
The code for the project has been primarily done in Jupyter Notebook and it saves and generate all the weights and plots upon execution.
After conducting the above experiments, it can be concluded that:
- Random forests technique (max_depth = 16) performed with 98-99% avg. accuracy
- This was followed by Neural Network of 2 hidden layers with 36 and 24 units respectively with 98% avg. accuracy. (LR = 0.001 and ReLU activation function)
- The 3rd best model was GDA with 92% average accuracy in our analysis.
- Anmol Kumar (2018382)
- Tejas Dubhir (2018110)
- Rishabh Chauhan (2018256)