Real-time recognition of a fall and rapid propagation of an emergency can minimize human casualties. This algorithm based on action recognition can detect when a person falls and trigger an alarm. It can be used for indoor and outdoor fall prevention purposes in hospitals, factories, construction sites, health clubs, and on the street. You can set it so that the alarm does not go off when there is movement in a lying position, like in a gym. It is designed to minimize alarms and enable detection under various lighting conditions during both day and night.