BTS-rPPG: Orthogonal Butterfly Temporal Shifting for Remote Photoplethysmography
This work addresses a domain-specific problem in contactless physiological sensing for healthcare or monitoring applications, representing an incremental advancement in temporal modeling for rPPG.
The paper tackled the challenge of limited temporal receptive fields in remote photoplethysmography (rPPG) by proposing BTS-rPPG, a framework using Orthogonal Butterfly Temporal Shifting, which improved long-range temporal modeling and outperformed existing methods on benchmark datasets.
Remote photoplethysmography (rPPG) enables contactless physiological sensing from facial videos by analyzing subtle appearance variations induced by blood circulation. However, modeling the temporal dynamics of these signals remains challenging, as many deep learning methods rely on temporal shifting or convolutional operators that aggregate information primarily from neighboring frames, resulting in predominantly local temporal modeling and limited temporal receptive fields. To address this limitation, we propose BTS-rPPG, a temporal modeling framework based on Orthogonal Butterfly Temporal Shifting (BTS). Inspired by the butterfly communication pattern in the Fast Fourier Transform (FFT), BTS establishes structured frame interactions via an XOR-based butterfly pairing schedule, progressively expanding the temporal receptive field and enabling efficient propagation of information across distant frames. Furthermore, we introduce an orthogonal feature transfer mechanism (OFT) that filters the source feature with respect to the target context before temporal shifting, retaining only the orthogonal component for cross-frame transmission. This reduces redundant feature propagation and encourages complementary temporal interaction. Extensive experiments on multiple benchmark datasets demonstrate that BTS-rPPG improves long-range temporal modeling of physiological dynamics and consistently outperforms existing temporal modeling strategies for rPPG estimation.