An elderly fall detection method and system based on deep learning

Abstract

A method and system for detecting falls of the elderly based on deep learning, including: collecting scene image data through a camera to determine a target detection area; according to the scene image data, calculating the instantaneous motion velocity field of all pixels in the current frame image with preset logic, Acquiring a pixel motion velocity field image; gathering similar pixels according to the pixel motion velocity field image and preset similarity conditions to form a candidate moving object area, obtaining candidate moving objects in the candidate moving object area; screening out new moving objects among the candidate moving objects; According to the deep learning method, the pedestrian target is identified from the new moving target; the target tracking list information of the pedestrian target is updated; based on the change state of the target tracking list information and the preset judgment conditions, it is judged whether the pedestrian target has fallen and an alarm is issued.

Description

technical field [0001] The invention relates to a pedestrian behavior detection method, in particular to a deep learning-based elderly fall detection method and system. Background technique [0002] As time goes by, China has gradually entered an aging society, coupled with changes in young people's life concepts and lifestyles, resulting in an increasing number of "empty nesters". Therefore, the accompanying "empty nesters" The safety of the "nest old man" has also become an issue of greater concern to the children. Due to the degradation of the physiological functions of the elderly, falls have become the primary issue for the health and safety of the elderly, and the reason why most falls will cause serious physical injury to the elderly is that the elderly have not been discovered in time after falling, thus missing the best time. Time for treatment. Therefore, developing a set of accurate and real-time fall detection technology for the elderly will surely produce huge...

AI Technical Summary

Problems solved by technology

The advantage of this type of method is non-contact detection and low cost. The disadvantage is that the background modeling is not accurate, and the extracted foreground block features have large errors, resulting in many false detections and missed detections.

In recent years, deep learning technology has been widely concerned and applied in various fields because it can simulate the neural network of the human brain and perform accurate nonlinear prediction. However, the disadvantage of this technology is that the model consumes a lot of memory and a large amount of calculation. , it is impossible to achieve real-time detection in the video environment. The reason is that the deep learning algorithm is aimed at a single static image and does not make full use of inter-frame correlation information, so it needs to consume a lot of memory and computing power

[0004] In the existing technology, it is necessary to carry wearable sensors with you. The cost of the equipment is high, and it is extremely inconvenient to use. The deep learning model consumes a lot of memory and has a large amount of calculation. It cannot be detected in real time in the video environment. The reason is that deep learning The algorithm is aimed at a single static image, and does not make full use of inter-frame correlation information. The disadvantage is that the background modeling is not accurate, and the extracted foreground block features have large errors, resulting in many false detections and missed detections. There are high hardware costs and computational complexity. Technical problems such as large space occupation, low information utilization rate and low accuracy of monitoring results

Images