Development of Various Machine Learning and Deep Learning models for Human Activity Recognition

Basketball Game Activities recognition (HAR) using Classical Machine Learning techniques and also Deep Learning algorithms and a brief comparison between them.

• Fused the data from e-sense and bangle.js sensors together after data cleaning, handling missing data, and resampling using interpolation.

• Implemented Sliding window technique and Stratified splitting technique and performed per-subject and cross-subject validations.

• Trained and worked on optimization of KNN, SVM, CNN, and LSTM networks.