Acoustic Fall Detection using Genetic K-NN and SVM
M Amsaprabhaa · 2023
Falls are abnormal movement coordination that forces a person to rest on the ground causing significant health risks involuntarily. This study aims to develop an Acoustic Fall Detection system based on the Genetic K-Nearest Neighbour (ADF-GKNN) feature selection approach and extract significant audio features from spectrogram and Mel Frequency Cepstral Coefficients (MFCCs). The support vector machine classifier is used to identify the falls using the extracted features. The audio signals are extracted from two benchmarked fall video datasets for experimentation. Classification results obtained using the proposed frameworks outperform other feature selection techniques regarding prediction accuracy with 90.6 and 91.8 percentages.