Automatic fall risk detection based on imbalanced data

Abstract

In recent years, the declining birthrate and ageing population have gradually brought countries into an ageing society. In regards to the accidents that occur amongst the elderly, falls are an important problem that quickly causes indirect physical loss. In this thesis, we propose a pose estimation-based fall detection algorithm to detect fall risks. Since fall data is rare in real-world situations, we train and evaluate our approach in a highly imbalanced data setting. We assess not only different imbalanced data handling methods, but also different machine learning algorithms. After oversampling on our training data, the K-Nearest Neighbors (KNN) algorithm achieves the best performance. This experiment provides evidence that our approach is more interpretable, with key features from skeleton information, and workable in multi-people scenarios.