Abstract: Falls are a significant contributor to unintentional fatalities and injuries on a global scale. People with balance impairments are particularly susceptible to falling. While surveys and lab simulations have provided valuable insights, there is a lack of prospective data on actual falls in real-world settings. Acquiring such information is crucial for identifying fall risks and designing effective systems for detecting and alerting falls. The technological areas of machine vision and ML are based on the ability of an individual to trace the actions of others. Action is a term used to describe a sequence of bodily functions that include several body parts working together at once. The comparison is conducted on any kind of remark against a predetermined pattern in machine vision, and the activity is recognized and labelled later on. The SVM is the classifier which is applied in the previous work for recognizing the activities of individuals. The SVM classifier performs poorly for identifying human actions. The presented work suggests a hybrid approach which is the combination of CNN and LSTM model. The proposed model achieves an accuracy of upto 87 percent for the human fall detection.

Fall Detection, SVM, CNN, LSTM, Sensors