BPM-Net: non-contact blood pressure measuring network based on face videos

Blood pressure indicates cardiac function and peripheral vascular resistance and is critical for disease diagnosis. Traditionally, blood pressure data are mainly acquired through contact sensors, which require high maintenance and may be inconvenient and unfriendly to some people (e.g., burn patients). In this paper, we proposed an efficient non-contact blood pressure measurement network based on face videos. First, an innovative oversampling training strategy is proposed to handle the unbalanced data distribution. The input video sequences are first normalized and converted to our proposed YUVT color space. Then the spatio-temporal slicer encodes it into a multi-domain spatio-temporal mapping. Finally, the feature extractor composed of a series backbone network and LSTM fits the high-dimensional feature, which is fed into blood pressure classifier to locates the blood pressure interval. The blood pressure calculator combines the results of the feature extractor and the blood pressure classifier to output the final blood pressure value. We tested BPM-Net on MMSE-HR dataset, the MAE of systolic blood pressure reached 12.35 mmHg and that of diastolic blood pressure reached 9.5 mmHg. Experimental results on MMSE-HR show that the network outperforms existing state-of-the-art methods.