Uloženo v:
| Hlavní autoři: | , |
|---|---|
| Médium: | Recurso digital |
| Jazyk: | |
| Vydáno: |
Zenodo
2026
|
| Témata: | |
| On-line přístup: | https://doi.org/10.5281/zenodo.19667695 |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
Obsah:
- <p>Among the elderly in the world, falls are considered to be the major causes of injuries, hospitalization, and death. Quick response can greatly decrease the number of medical complications as the fall incidents can be detected in time. Conventional fall detection systems use wearable sensors that are usually awkward or obtrusive and not always convenient to use in their entirety. The proposed research paper suggests a non-invasive fall detection system of the elderly using data provided by cameras and implements MediaPipe-based human pose estimation, OpenCV-velocity of movement, angles and temporal shifts of the posture to enhance the rate of detection and false alarms. The system will incorporate an automated alert system to alert the caregivers or medical staff members in real time when a fall takes place. Experimental evidence shows that the method proposed is highly accurate, sensitive and robust in different environmental conditions which make it applicable in real world application of smart home and assisted living facility.</p>