The study evaluates explainable AI methods for Remote Photoplethysmography (rPPG) using the RhythmFormer model across two datasets (NCKU-rPPG and UBFC-rPPG). Quantitative metrics—skin coverage and the Salience-guided Faithfulness Coefficient (SaCo)—were applied to four explanation techniques (raw attention, rollout, attention flow, and Beyond Intuition). Beyond Intuition consistently achieved the highest coverage and SaCo, while other methods showed weak or opposite correlations with performance measures, and performance degraded notably under low illumination (40 lux).
arXiv:2606. 13839v1 Announce Type: cross Abstract: Remote photoplethysmography (rPPG) transformers achieve low heart-rate error on benchmarks, yet their decisions remain opaque--a growing concern as rPPG moves toward clinical heart rate estimation.
The paper introduces Heartian, a physiology‑aware framework that augments Gaussian head avatars with cardiac‑cycle‑dependent albedo modulation, enabling the encoding of remote photoplethysmography (rPPG) signals. By supervising with synchronized contact PPG, the method models the cardiac waveform as a sum of two Gaussians and learns per‑frame spatial residuals via a lightweight MLP. Experiments on 152 stationary recordings from UBFC‑rPPG, PURE, and MMPD show heart‑rate estimation errors as low as 0.29 bpm MAE and 0.38 % MAPE, while preserving reconstruction quality with negligible PSNR loss.
The paper investigates 12‑class body‑only emotion recognition from skeleton motion using a leave‑performer‑out evaluation, where chance accuracy is 8.3% and a reproduced STGCN++ baseline scores 25.73% Macro‑F1. By ensembling eleven models with orthogonal error modes, the authors achieve 36.80% Macro‑F1, a 43% relative improvement over the baseline. They also introduce a tested explanation suite that demonstrates the ensemble’s decisions rely on motion‑grounded body‑region evidence, aligning strongly with Laban Movement Analysis attributes rather than classical kinematics, while showing diffuse temporal saliency.
arXiv:2607. 13738v2 Announce Type: replace-cross Abstract: Deep video models estimate left-ventricular ejection fraction (EF) from echocardiography with near-expert accuracy, and post-hoc attribution is increasingly used to certify that such models look at the right place.
The study examined how well specific physiological properties of contact photoplethysmography (PPG) can be recovered when converted to camera-derived remote photoplethysmography (rPPG) across 655 recordings. Using the CHROM method, the authors found that while overall heart‑rate accuracy was modest, many dynamic properties (e.g., autocorrelation, spectral measures, Lyapunov exponents) showed little recording‑specific correspondence, and differences varied with skin tone and observation conditions. The results demonstrate that preserving individual recording characteristics depends on the property and conditions, and population‑level agreement does not guarantee individual‑level fidelity.
Gaussian head avatars typically model intrinsic facial appearance as temporally static, omitting subtle cardiac-induced skin-color variation. We propose $\unicode{x1F493}$Heartian, a physiology-aware...
The study examined how well physiological properties from contact photoplethysmography (PPG) can be recovered by camera-based remote photoplethysmography (rPPG) across 655 recordings. Using the CHROM method, the authors found that while heart‑rate estimates showed modest accuracy, other dynamic measures such as autocorrelation, spectral content, and Lyapunov exponents largely failed to preserve recording‑specific characteristics. They also observed that lighting and motion influence endpoint accuracy, yet these factors did not explain the lack of dynamical correspondence.
arXiv:2607. 13738v1 Announce Type: cross Abstract: Background and Objective: Deep video models estimate left-ventricular ejection fraction (EF) from echocardiography with near-expert accuracy, and post-hoc attribution (Chefer relevance for transformers, Grad-CAM for CNNs) is increasingly used to certify that models "look at the right place.
The paper evaluates six onboarding strategies for federated wearable models on five datasets using a leakage‑controlled protocol that fixes source checkpoints and separates calibration from evaluation. Results show that while average accuracy is high, person‑level performance can drop significantly, with some methods causing negative transfer for certain users. The study highlights that mean accuracy alone is insufficient and provides an auditable benchmark and failure map for future development.
arXiv:2609.05522v1 Announce Type: cross Abstract: Eye-tracking data are expensive to collect, requiring specialized hardware and controlled laboratory conditions, and difficult to share because of pr...
arXiv:2607. 01400v1 Announce Type: cross Abstract: Deep multimodal brain-encoding models now predict fMRI responses to naturalistic video with high accuracy.
arXiv:2609.16551v1 Announce Type: new Abstract: Self-supervised learning (SSL) can reduce the need for labelled medical images, but the choice of pretext objective remains unclear for lung ultrasound...