一种基于双流网络的跌倒检测算法

A fall detection algorithm based on dual flow network

跌倒动作识别在独居老人的监护与医疗领域有着巨大的前景与应用价值。随着计算机视觉与深度学习的不断发展,动作识别技术也有着显著的近展开,文章针对通过双流网络与LSTM网络各自的优势进行结合,通过建立40人的室内动作数据集对模型的训练。结果表明,时序双流网络增强了模型在时序方向上的表达能力,提高了模型的检测准确率,在独居老人的医疗监护方面有一定的适用价值。

Fall movement recognition has great prospect and application value in the field of monitoring and medical treatment for the elderly living alone. With the continuous development of computer vision and deep learning, motion recognition technology also has a significant near development. This paper aims at the combination of the advantages of dual flow network and LSTM network, through the establishment of 40 people’s indoor motion data set to train the model. The results show that the time-series dual flow network enhances the expression ability of the model in the time-series direction, improves the detection accuracy of the model, and has a certain applicable value in the medical care of the elderly living alone.