Fall detection alarm method based on embedded edge artificial intelligence

Abstract

The invention relates to the technical field of intelligent wearable devices of the Internet of Things, in particular to a fall detection alarm method based on embedded edge artificial intelligence, which comprises the following steps: S0, establishing a fall detection alarm system based on embedded edge artificial intelligence; S1, creating a characteristic value data set; S2, creating an FDMLP model; S3, performing fall detection; and S4, alarming and asking for help. According to the method, an optimization method that the learning rate is subjected to exponential attenuation according to the number of iterations is adopted; the characteristic value data set for old people fall detection is established; full integer quantization is performed on the model, and successfully operated on embedded equipment; state information and positioning information are sent to a cloud server through a communication module and a positioning module, and data are synchronized to a mobile terminal application program of a family member for rescue in time. The operation speed of the model is effectively improved and the storage consumption of the model is reduced on the premise of ensuring high-precision judgment; and all identification activities are realized at the equipment end, so that the privacy data of the user is protected.

Description

technical field [0001] The invention relates to the technical field of smart wearable devices of the Internet of Things, in particular to a fall detection and alarm method based on embedded edge artificial intelligence. Background technique [0002] According to the World Health Organization, the global elderly population will reach 1.2 billion by 2025. With the aging of the society and the increasingly prominent problems of empty-nest families, the hospitalization rate and death rate of the elderly due to falls are rising, which has become a difficult problem for children to take care of the elderly. The field of nursing care for the elderly in my country needs a more scientific and efficient nursing model and a complete pension system. Under the condition of not hindering the normal life of the elderly, scientific and effective methods should be used to make the fallen get rescued in time. [0003] At present, a variety of artificial intelligence-based fall detection sch...

AI Technical Summary

Problems solved by technology

[0004] CNN has the characteristics of shared convolution kernel, so it has advantages in processing high-dimensional data. It is usually used for visual recognition, but due to the encapsulation of feature extraction, it is not easy to improve network performance; LSTM is convenient for sequence modeling, but when the sequence length exceeds a certain limit , there will be a vanishing gradient problem

The traditional video-based detection algorithm is relatively mature, but it will violate the privacy of the elderly, and is easily affected by the environment and light, and the camera can only be fixed at a specific position; the application scene of the method based on environmental variables is limited, and the deployment cost is high; wearable Device detection method The detection device is integrated with the mobile terminal, making full use of the perception function, computing function and communication function of the embedded edge artificial intelligence, but the portability of the wearable detection device needs to be improved

[0005] Therefore, this application has designed and developed a fall detection and alarm method based on embedded edge artificial intelligence, aiming to help the elderly who go out alone, and solve the plight of the target population who cannot get timely assistance because their family members are not around when they accidentally fall

Images