Drone Assisted Remote Wellness Monitoring Using RGB Camera
14th International Symposium on Digital Forensics and Security, ISDFS 2026, Massachusetts, Amerika Birleşik Devletleri, 19 - 20 Mart 2026, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Doi Numarası: 10.1109/isdfs69419.2026.11459053
- Basıldığı Şehir: Massachusetts
- Basıldığı Ülke: Amerika Birleşik Devletleri
- Anahtar Kelimeler: Disaster Response Telemedicine, Drone-Assisted Healthcare, Edge AI, Heart Rate Estimation, Quality Gating, Refusal Policy, Remote Patient Monitoring, rPPG
- Maltepe Üniversitesi Adresli: Evet
Özet
In remote rural regions and severe weather conditions (e.g., snowstorms and floods), emergency response teams may be delayed or unable to reach patients in time. This motivates drone-assisted remote triage workflows where a firstresponse UAV provides early situational and physiological telemetry to a command center. We present a practical RGB camerabased wellness monitoring pipeline that estimates physiological indicators from video via remote photoplethysmography (rPPG) and motion-derived signals. Unlike accuracy-only approaches, the proposed system targets safety-critical operation by integrating explicit quality gating and a refusal policy to avoid wrong-but-confident outputs under adverse capture conditions (motion, unstable sampling, intermittent face loss, or illumination drift). The pipeline consists of face ROI extraction, POS-inspired signal projection, temporal stabilization, and an interpretable Signal Quality Score (SQS) used for selective publishing. In a pilot real-time capture log (N=3448 JSONL rows; effective median capture rate ≈ 3.72 FPS), the system achieves high heartrate telemetry availability (97.71% coverage) while transparently abstaining on insufficient-evidence segments with explicit reason codes. Since the pilot operates at substantially lower frame rates than typical rPPG settings, we emphasize safety-first availability under strict gating rather than clinical-grade accuracy, and conservatively disable HRV reporting under low-FPS conditions. Failure-mode analysis indicates that dominant rejections are driven by sampling instability and missing respiration evidence. The proposed design supports drone-enabled telehealth triage, where reliability, interpretability, and honest abstention are critical for operational decision-making.