Professional Research
Peer-reviewed and conference-submitted work in computer vision and edge AI.
IEEE DG 2026
2026 — VidOpt LLC / ITKAN Institute
Adaptive Detection Scheduling for Multi-Object Tracking on Resource-Constrained Edge Devices
A systematic study of detection scheduling for a YOLOv8n detector coupled with a SORT tracker, reaching a 16× speedup while mapping where Kalman coasting breaks down — plus an honest negative result on a content-adaptive scheduler.
View Paper (PDF) →
IEEE BigData 2026 · HS Symposium
2026 — Simulation Study
Multimodal Confirmation for Reducing False Fall Alerts: A Simulation Study of the ResQ Decision Policy
A Monte Carlo evaluation of 100,000 monitoring events across 27 operating conditions, cutting false emergency alerts by a mean of 95.9% and deriving the closed-form behavior that decouples perception from human confirmation.
View Paper (PDF) →