Dataset and Code: Benchmarking rPPG on Fitzpatrick IV–V Skin — A 100-Subject South Indian Smartphone Dataset

Companion dataset, results, and code for the TENCON 2026 paper "Benchmarking rPPG on Fitzpatrick IV–V Skin: A 100-Subject South Indian Smartphone Dataset with Classical and Domain-Adaptive Deep Learning Evaluation." Includes per-subject demographic and ground-truth HR/SpO2 labels, classical rPPG algorithm results (GREEN/CHROM/POS), deep-learning (temporal CNN, LOSO cross-validated) results, analysis code, and the camera-ready manuscript. Raw facial video is not included; available on request under a data use agreement (see README).