2026-06-26T11:48:43.625Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:11409 2025-06-13T13:58:39Z 3

YOLOv8 Model Architecture Selection for Human Fall Detection, Chapter in LNNS Lecture Notes in Networks and Systems: ICDAM 2024: International Conference on Data Analytics & Management, Springer, volume 1301 2025 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/11409 https://link.springer.com/chapter/10.1007/978-981-96-3372-2_15 T. Zivkovic M. Zivkovic L. Jovanovic J. Kaljevic M. Dobrojevic N. Bacanin Detecting human falls is vital in several domains, from the elderly care facilities to the dangerous workplaces. Falls in case of elderly people may result in serious injuries, where prompt assistance is necessary to reduce the risk of heavier health problems and complications. The application of fall detection systems may result in significant cost reduction, as falls may lead to hospitalization, surgery, and lengthy rehabilitation. This paper investigates the performance of YOLOv8 models for this task. These models are affordable, their training is fast and cheap, and they are recognized of their speed and efficient operation. The simulation outcomes suggest that these affordable models are very suitable for the fall detection task. bookPart Springer, Singapore 219 227 10.1007/978-981-96-3372-2_15 LNNS Lecture Notes in Networks and Systems: ICDAM 2024: Proceedings of Data Analytics and Management, volume 1301