English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Recent advances in supervised deep learning methods are enabling remotemeasurements of photoplethysmography-based physiological signals using facialvideos. The performance of these supervised methods, however, are dependent onthe availability of large labelled data. Contrastive learning as aself-supervised method has recently achieved state-of-the-art performances inlearning representative data features by maximising mutual information betweendifferent augmented views. However, existing data augmentation techniques forcontrastive learning are not designed to learn physiological signals fromvideos and often fail when there are complicated noise and subtle and periodiccolour or shape variations between video frames. To address these problems, wepresent a novel self-supervised spatiotemporal learning framework for remotephysiological signal representation learning, where there is a lack of labelledtraining data. Firstly, we propose a landmark-based spatial augmentation thatsplits the face into several informative parts based on the Shafer dichromaticreflection model to characterise subtle skin colour fluctuations. We alsoformulate a sparsity-based temporal augmentation exploiting Nyquist-Shannonsampling theorem to effectively capture periodic temporal changes by modellingphysiological signal features. Furthermore, we introduce a constrainedspatiotemporal loss which generates pseudo-labels for augmented video clips. Itis used to regulate the training process and handle complicated noise. Weevaluated our framework on 3 public datasets and demonstrated superiorperformances than other self-supervised methods and achieved competitiveaccuracy compared to the state-of-the-art supervised methods.

Self-supervised Representation Learning Framework for Remote Physiological Measurement Using Spatiotemporal Augmentation Loss