Adaptive Frame Sampling for Real-Time Video Object Detection and Multi-Object Tracking on Edge Devices
Journal of Information and Communications Technology: Algorithms, Systems and Applications · 30.06.2026 · 2 (2)
DOI: 10.64189/ict.26310 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Information and Communications Technology: Algorithms, Systems and Applications, online erschienen am 30.06.2026. Verfasst von Pritee A. Parwekar und 1 weiteren Autor:innen.
Abstract (Originalsprache)
Real-time multi-object tracking on CPU-only edge devices is constrained by the high per-frame inference cost of deep neural network detectors. We present the Adaptive Frame Sampling System (AFSS), a training-free, architecture-agnostic framework that dynamically allocates computation across three actions per frame: full YOLOv8 inference (FULL), phase-correlation feature warping (WARP), or result reuse (SKIP), governed by a lightweight scene complexity estimator (
Zitieren
APA
Parwekar, P. A., & Kadiri, A. (2026). Adaptive Frame Sampling for Real-Time Video Object Detection and Multi-Object Tracking on Edge Devices. Journal of Information and Communications Technology: Algorithms, Systems and Applications, 2(2). https://doi.org/10.64189/ict.26310
Vancouver
Parwekar PA, Kadiri A. Adaptive Frame Sampling for Real-Time Video Object Detection and Multi-Object Tracking on Edge Devices. Journal of Information and Communications Technology: Algorithms, Systems and Applications. 2026;2(2). doi:10.64189/ict.26310
BibTeX
@article{Parwekar2026,
author = {Parwekar, Pritee A. and Kadiri, Adarsh},
title = {Adaptive Frame Sampling for Real-Time Video Object Detection and Multi-Object Tracking on Edge Devices},
journal = {Journal of Information and Communications Technology: Algorithms, Systems and Applications},
year = {2026},
volume = {2},
number = {2},
doi = {10.64189/ict.26310}
}
EUVIMED