A Deep Learning approach for detection and analysis of respiratory infections in covid-19 patients using RGB and infrared images.

Within a short period, the severe acute respiratory syndrome Coronavirus Disease 2019 (COVID-19) has become a devastating global pandemic, causing enormous losses to human civilization worldwide. A significant feature of COVID-19, according to recent investigations, is an altered respiratory state induced by viral infections. In this paper, we present a non-contact method for screening the respiratory health of COVID-19 patients using RGB-infrared sensors to analyze their breathing patterns. The block diagram the proposed method is shown in Fig. 1. First, we use facial recognition to obtain breathing data from the individuals. The respiratory data is applied to multiple neural networks, including LSTM, BiLSTM, GRU, and BiGRU. An attention mechanism is then used in the neural network to obtain a health screening result from the respiration dataset. With an accuracy of 70.83 percent, our BiGRU model accurately identifies the respiratory health condition whether it is normal or abnormal.