Fall detection method and system

Abstract

The invention provides a fall detection method, which includes two stages of off-line training and on-line detection. In the offline training stage, multiple sub-classifiers are trained by combining random sample selection and random classification attribute selection; in the online fall detection stage, the final judgment result is given based on the classification results of multiple sub-classifiers. Experiments show that the fall detection method provided by the present invention can obtain an accuracy rate of 95.2%, a sensitivity of 90.6% and a specificity of 93.5%, which is obviously better than the fall detection method based on SVM and BP neural network, and can more accurately detect the fall behavior, with Strong generalization ability and robustness.

Description

technical field [0001] The invention belongs to the field of detection and intelligent monitoring, and in particular relates to a fall detection method and system. Background technique [0002] With the increasing population aging and the increasing number of empty-nesters, the problem of personal injury caused by falls of the elderly has become increasingly prominent. According to statistics, falls have become the fourth leading cause of injury death in my country, and the first cause of injury death in the elderly over 65 years old, and the fall mortality rate further increases with the increase of the age of the elderly. In addition to directly causing the death of the elderly, falls also lead to a large number of disabilities, reduce the activity and range of activities of the elderly, and seriously affect the quality of life and physical and mental health of the elderly. [0003] In recent years, in order to reduce the further injury caused by accidental falls to the e...

AI Technical Summary

Problems solved by technology

However, due to the scarcity of real elderly fall data and the inability to use the elderly for fall data collection, most research work uses young people to simulate and collect experimental data sets for testing. Due to the small sample size and low representativeness, it often leads to classification. The device is over-fitting, and the accuracy and robustness of fall detection are not high

Images