Automatic Fall Risk Detection Based on Imbalanced Data

Abstract

In recent years, the declining birthrate and aging population have gradually brought countries into an ageing society. Regarding accidents that occur amongst the elderly, falls are an essential problem that quickly causes indirect physical loss. In this paper, we propose a pose estimation-based fall detection algorithm to detect fall risks. We use body ratio, acceleration and deflection as key features instead of using the body keypoints coordinates. 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. The F1 scores for three different classes, Normal, Fall, and Lying, are 1.00, 0.85 and 0.96, which is comparable to previous research. The experiment shows that our approach is more interpretable with the key feature from skeleton information. Moreover, it can apply in multi-people scenarios and has robustness on medium occlusion.

Advances in Skeleton-Based Fall Detection in RGB Videos: From Handcrafted to Deep Learning Approaches

“…Distance metrics include Euclidean distance, Manhattan distance, etc. As a classification technique, KNN has been employed in several reviewed studies [39], [41]- [43].…”

“…The SVM can be extended to the kernelized SVM in the case of non-linear data, which uses the kernel function to shift the original data into a higherdimensional space where the data can be linearly separated. There have been numerous reviewed studies using SVM for classification in skeleton-based fall detection systems [39], [41]- [43], [46].…”

“…Boosting algorithms include AdaBoost [49], XGBoost [50], etc. In the reviewed work [43], AdaBoost and XGBoost were applied during the classification phase.…”

“…Skeleton data is considered invalid if it belongs to a person who is not involved in the fall event [41], [62], [46], or if it lacks specific keypoints such as the neck, hip keypoints [61], [63], or if it contains wrong keypoint data such as the detection of other objects, such as a chair, as a human body [66]. Some keypoints that do not contribute to fall detection can be removed from the whole detected keypoint sets, as argued by the authors in [43], [58], [59], [67]. For example, the authors in [58] used only 8 keypoints (R.Knee, L.Knee, R.Hip, Mid.Hip, L.Hip, R.Shoulder, L.Shoulder, Neck -see Fig.…”

Fall detection based on OpenPose and MobileNetV2 network

“…Head position, body aspect ratio parameters and Motion History Image (MHI) were used to represent human posture information for detecting falls [7]. What's more, the authors used body proportion, acceleration and deflection as key features of the human body to detect falls [8]. And the fall motion vector modeling was used for falling detection [9].…”