PreFallKD: Pre-Impact Fall Detection via CNN-ViT Knowledge Distillation
This work addresses fall prevention for elderly individuals using wearable systems, offering an incremental improvement by balancing detection performance with computational efficiency for resource-constrained devices.
The paper tackles pre-impact fall detection for wearable devices by proposing PreFallKD, a knowledge distillation method that transfers detection knowledge from a vision transformer teacher to a lightweight CNN student, achieving an F1-score of 92.66% and a lead time of 551.3 ms on the KFall dataset.
Fall accidents are critical issues in an aging and aged society. Recently, many researchers developed pre-impact fall detection systems using deep learning to support wearable-based fall protection systems for preventing severe injuries. However, most works only employed simple neural network models instead of complex models considering the usability in resource-constrained mobile devices and strict latency requirements. In this work, we propose a novel pre-impact fall detection via CNN-ViT knowledge distillation, namely PreFallKD, to strike a balance between detection performance and computational complexity. The proposed PreFallKD transfers the detection knowledge from the pre-trained teacher model (vision transformer) to the student model (lightweight convolutional neural networks). Additionally, we apply data augmentation techniques to tackle issues of data imbalance. We conduct the experiment on the KFall public dataset and compare PreFallKD with other state-of-the-art models. The experiment results show that PreFallKD could boost the student model during the testing phase and achieves reliable F1-score (92.66%) and lead time (551.3 ms).