Remote Photoplethysmography (rPPG) is a non-contact method for measuring heart rate (HR) through facial video, breaking the constraints of contact measurements and offering broad application prospects. However, in real monitoring scenarios, the distance of subjects and facial illumination often vary. This leads to degradation of rPPG signals due to changes in facial resolution and light intensity. This paper presents the High and Low Frequency Feature Adversarial Learning Network (HLFF-AL), which utilizes feature space interpolation and composite feature capture to perform multi-band global and local signal extraction. By adopting an adversarial learning strategy, it enhances the capability to capture semantic information of rPPG under varying resolutions and lighting, bridging the differences in rPPG signal generation due to resolution changes. While maintaining accuracy in HR measuring, the robustness of the network is improved to achieve robust measurement of HR via rPPG. Even under high-resolution conditions, the Mean Absolute Error (MAE) for the three public datasets—unchanging lighting (COHFACE, UBFC-rPPG) and varying lighting (BUAA-MIHR)—reached 1.10 bpm, 1.64 bpm, and 4.41 bpm, respectively, with a mere 0.29 bpm difference in MAE between high and low resolutions on the BUAA-MIHR dataset. This demonstrates that HLFF-AL can predict more robust rPPG signals in various resolutions and lighting scenarios, achieving competitive results compared to state-of-the-art methods.

Heart rate (HR) is an important indicator reflecting the overall physical and mental health of the human body, playing a crucial role in diagnosing cardiovascular and neurological diseases. Recent research has revealed that variations in the light absorption of human skin captured through facial video over the cardiac cycle, due to changes in blood volume, can be utilized for non-contact HR estimation. However, most existing methods rely on single-modal video sources (such as RGB or NIR), which often yield suboptimal results due to noise and the limitations of a single information source. To overcome these challenges, this paper proposes a multimodal information fusion architecture named the spatiotemporal sensitive network (SS-Net) for non-contact heart rate estimation. Firstly, spatiotemporal feature maps are utilized to extract physiological signals from RGB and NIR videos effectively. Next, a spatiotemporal sensitive (SS) module is introduced to extract useful physiological signal information from both RGB and NIR spatiotemporal maps. Finally, a multi-level spatiotemporal context fusion (MLSC) module is designed to fuse and complement information between the visible light and infrared modalities. Then, different levels of fused features are refined in task-specific branches to predict both remote photoplethysmography (rPPG) signals and heart rate (HR) signals. Experiments conducted on three datasets demonstrate that the proposed SS-Net achieves superior performance compared to existing methods.

Remote photoplethysmography (rPPG) enables noncontact estimation of vital signs, particularly heart rate, by analyzing subtle periodic skin color variations in facial videos. While deep learning has advanced rPPG signal extraction, existing methods rely on carefully designed task-specific architectures that are costly to develop and generalize poorly. In this work, we demonstrate that the general video foundation model VideoMAE v2 can be effectively adapted to the rPPG signal regression task by introducing only a lightweight adapter, without modifying its pretrained backbone. We freeze the entire VideoMAE v2 encoder and introduce a Temporal Difference Convolutional Adapter to capture the subtle interframe intensity differences. To address the mismatch between VideoMAE v2′s short input window (16 frames) and the long temporal context typically required for robust rPPG extraction (e.g., 160 frames), we adopt an overlapping sliding window strategy for segmented inference and reconstruct the full signal through weighted temporal aggregation. On the COHFACE and UBFC-rPPG datasets, our method achieves mean absolute errors (MAEs) of 0.90 and 1.55, reducing the error by more than 55% and 42%, respectively, compared to PhysFormer (2.00 and 2.70). Furthermore, on challenging real-world datasets such as BUAA-MIHR, which features strong illumination variations, and VIPL-HR, which involves significant head movements, our approach achieves MAEs of 6.68 and 8.23, respectively, despite incorporating no task-specific robustness modules. These results demonstrate stable rPPG signal recovery and validate the feasibility of leveraging general video foundation models for physiological signal perception.