Fall detection mostly depends on sensor equipment.The method is highly influenced by equip-ment and environmental factors,and often can not work well.In addition,vision-based methods are often not effective in terms of real-time and robust.In order to solve these problems,a lightweight fall detec-tion algorithm is proposed with strong robustness and convenient deployment in embedded devices.Tak-ing YOLOv5 as the benchmark model,the lightweight attention mechanism module is firstly integrated to make the network focus on the target area to be identified and enhance the recognition accuracy of the network.Secondly,the model is pruned by the model compression method,which reduces the volume and calculation.Therefore it makes the model lightweight,so as to improve the reasoning speed and facilitate deployment in embedded devices.Finally,knowledge distillation is carried out on the pruned model,which can improve the detection accuracy without increasing the complexity of the model.The experimental re-sults show that compared with the benchmark model,the mAP of this model is increased by 1.7％,the recall is increased by 1.2％,the model volume is reduced by 79.1％,and the floating-point operation is reduced by 70.9％.The proposed model is deployed on the embedded device Jetson Nano,and the detec-tion speed is up to 13.2 frame/s,which basically meets the requirements of real-time fall detection.
关键词
摔倒检测/注意力机制/轻量化/模型压缩/知识蒸馏
Key words
fall detection/attention mechanism/lightweight/model compression/knowledge distillation
维普
万方数据