Real-time activity and fall detection using transformer-based deep learning models for elderly care applications

View Full Paper

Ask AI

Bookmark

Authors

Discussion

This analysis demonstrates a transformer model improves fall detection and activity recognition, suggesting reliable applications for elderly care.

The transformer model achieved over 98% accuracy in real-time fall detection and activity recognition.
Evaluation on the MobiAct dataset showed improved performance over CNN-LSTM and temporal convolutional networks.
Sliding window segmentation was used to process data from wearable sensors, enhancing accuracy and applicability.
This research highlights the potential for real-time deployment using transformer-based solutions in elderly care.

Zafar et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark

This analysis demonstrates a transformer model improves fall detection and activity recognition, suggesting reliable applications for elderly care.