Akademik Veri Yönetim Sistemi
7th International Symposium on Innovative Approaches in Smart Technologies (ISAS 2023), İstanbul, Türkiye, 23 Kasım 2023
— Falls are a common occurrence, particularly among
the elderly, often leading to unintentional injuries and accidents that
can have severe consequences, necessitating swift intervention.
Individuals living alone frequently face delays in obtaining the
essential emergency medical assistance they require during such
incidents. Consequently, accurately detecting falls and promptly
initiating emergency aid is of paramount importance. In this work,
a fall detection system was developed using machine learning
algorithms, utilizing data collected from the MPU6050 IMU sensor
through the Edge-Impulse platform. This system automatically
identifies when a person experiences a fall and promptly notifies the
user or relevant individuals. The primary goal of this project is to
enhance the safety of elderly and injured individuals by enabling
rapid intervention during fall incidents. The system leverages Edge
Impulse to streamline the development of machine learning
algorithms. A dataset containing accelerometer signal recordings of
five normal activities and four falling cases is constructed. A
sequential neural network model was trained and accuracy of
88.29% on the test set was obtained. Further enhancements in the
accuracy of the real-time system can be achieved by increasing the
size of the dataset.