12 February 2021 in IEEE Transactions on Instrumentation and Measurement
Remote photoplethysmography (rPPG) technology is widely used to measure heart rate (HR) from facial video. However, the accuracy of rPPG signal extraction is affected by the slow content changes in long-range facial videos, and the extraction process is also susceptible to interference from illumination variation and head movement noise. To address the above problems, in this article, we propose an end-to-end effective time-domain attention network (ETA-rPPGNet). First, in order to overcome video redundancy information, we construct the time-domain segment subnet. The video is divided into several segments, which are fed into the subspace networks to extract important spatial facial features and aggregate temporal information, respectively. Then, the time-domain attention mechanism is designed in the backbone net. In this mechanism, the 1-D convolution is used to effectively model the information association in the local time domain, so as to improve the antinoise ability of the model. Finally, via the two-part loss function, the model can reduce the interference of other physiological signals. We conduct extensive experiments to verify the effectiveness of our model on the public PURE, COHFACE, UBFC-rPPG, and MMSE-HR data sets. Compared with other models, our model shows better performance in the measurement accuracy.
26 February 2021 in IEEE Transactions on Cognitive and Developmental Systems
In order to solve the problems of highly redundant spatial information and motion noise in the heart rate (HR) estimation from facial videos based on remote photoplethysmography (rPPG), this article proposes a novel HR estimation method based on spatial–temporal attention model. First, to reduce the redundant information and strengthen the association relationships of long-range videos, the spatial–temporal facial features are extracted by the 2-D convolutional neural network (2DCNN) and 3-D convolutional neural network (3DCNN), respectively. The aggregation function is adopted to incorporate feature maps into short segment spatial–temporal feature maps. Second, the spatial–temporal strip pooling is designed in the spatial–temporal attention module to reduce head movement noises. Then, via the two-part loss function, the model can focus more on the rPPG signal rather than the interference. We conduct extensive experiments on two public data sets to verify the effectiveness of our model. The experimental results show that the proposed method achieves significantly better performances than the state-of-the-art baselines: The mean absolute error could be reduced by 11% on the PURE data set, and by 25% on the COHFACE data set.
29 November 2021 in IEEE Transactions on Cognitive and Developmental Systems
Remote photoplethysmography (rPPG) based on computer vision technology is widely used to calculate the heart rate (HR) from facial videos. Existing rPPG techniques have been subject to some limitations (e.g., highly redundant spatial information, head movement noise and region of interest (ROI) selection). To address these limitations, this paper introduces an effective spatial-temporal attention network. A temporal fusion module is firstly proposed to fully exploit the time-domain information, aiming to reduce the redundant video information and strengthen the association relationships of long-range videos. Furthermore, a spatial attention mechanism is designed in the backbone net to precisely target the skin ROIs. Finally, to assist the network in learning the weights between channels, we project the RGB images using plane orthogonal to skin (POS) algorithm and add motion representation to complement physiological signals’ extraction. Extensive experiments on the public PURE, MMSE-HR, and UBFC-rPPG datasets demonstrate that our model achieve competitive results compared with other methods.