Deep learning based cough detection camera using enhanced features
作者：
Highlights：
• Deep learning model was developed for cough detection useful in pandemic situation.
• The detected cough sound is localized and visualized by sound camera in real-time.
• Novel velocity and acceleration features are used to detect cough sound effectively.
• Data augmentation was done to alleviated background noises and class imbalances.
• Trained model achieved F1 score of 90.0% with MFCC-V-A feature in a pilot test.
摘要
•Deep learning model was developed for cough detection useful in pandemic situation.•The detected cough sound is localized and visualized by sound camera in real-time.•Novel velocity and acceleration features are used to detect cough sound effectively.•Data augmentation was done to alleviated background noises and class imbalances.•Trained model achieved F1 score of 90.0% with MFCC-V-A feature in a pilot test.
论文关键词：Cough detection,Coronavirus,COVID-19,Deep learning,Feature engineering,Sound visualization
论文评审过程：Received 28 June 2021, Revised 24 May 2022, Accepted 6 June 2022, Available online 9 June 2022, Version of Record 15 June 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.117811
