Vision Based Fall Detection with BRISK Feature Descriptor
Swati Shilaskar, Shripad S. Bhatlawande, Ranveer Chavare, Rushikesh Joshi, Sakshi Jaiswal · 2023
Falls are the major issues which occurs unexpectedly. This can also lead sometimes to death. Elderly persons are at the top most risk and their life can also be in danger. For that purpose, proper fall detection system is needed to be developed which can detect the fall as soon as the fall event takes place. To make the system accurate, pre-processing of the images from the dataset is needed to be done properly. When a person is alone and the fall takes place, there is no one to detect the fall if the fall is major, the corresponding person will not be in the state to ask for help, in that situation this system can be used. Different machine learning classifiers namely Decision Tree, Random Forest Algorithm (RF) and KNN are used to determine the accuracy of the system. All these classifiers provide the accuracy on the basis of key points which are obtained with the help of BRISK feature descriptor. Different classifiers provided different accuracy among those classifiers Random Forest classifier provided the most highest accuracy of 75.30%.