Vision-Based Triage Model for Fall Detection

Artificial Intelligent (AI)has different methods to utilize technology in various scenarios. AI also has major impacts and values over emergency services. It can take actions and provide the ability to handle and support emergency cases that take place within a proper time to reduce the effect on the human lives. The paper addresses an AI model that forecasts and estimates syncope scenarios and incidents that may occur ahead of time or on the fly which results in immediate interventions and support to those cases to reduce complication. In this study, we have built a model using TensorFlow (TF) and Posenet platforms for pose estimation. Within this model, a dataset of 7,000 images for fainting incidents has been split into 80% for training and 20% for testing. The results of the model have shown a precision of 97% among those cases. The model might be used for future enhancements and upgrades providing more analytics on the level of fainting, and the results expected based on the person falling to the ground.