Federated Self-Supervised Learning for Security-Oriented CCTV Camera-Hardware Analysis
Journal of Eye-Innovation in Security Analysis · 12.08.2026 · 1 (20261) · 18
DOI: 10.68990/jiesa.202612 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Eye-Innovation in Security Analysis, online erschienen am 12.08.2026. Verfasst von Nadhem Ebrahim und 1 weiteren Autor:innen. Die Arbeit nennt 11 Literaturangaben.
Abstract (Originalsprache)
This study audits the Surveillance Camera Image Dataset (CCTV Cameras). We propose an honest federated self-supervised learning (FSSL) framework for privacy-aware analysis of surveillance-camera hardware images. The audited release contains 1,421 decodable images. No labels, bounding boxes, masks, class directories or predefined splits. Therefore, no supervised federated-learning model was trained, and this article reports no accuracy, F1, mAP, convergence, or confusion-matrix results. Instead, the audit establishes technical feasibility: 1,202 files decode as JPEG, 212 as WEBP, and 7 as PNG, although every filename ends in .jpg. There are approximately 1,183 square-like images, with widths ranging from 120 to 4,000 pixels and heights ranging from 120 to 5,184 pixels, and 18 visually verified near-duplicate components containing 37 images. We therefore propose a five-client non-IID protocol where content-normalized images are partitioned by source/domain and task-relevant embeddings, clients optimize a two-view self-supervised representation objective, and a coordinator applies FedAvg or FedProx through secure aggregation. The contribution produced is a reproducible dataset-feasibility study, leakage-aware annotation plan, and a testable future evaluation protocol for camera-infrastructure inventory and investigative asset triage – not crime, behavior, or identity inference.
Zitieren
APA
Ebrahim, N., & Nair, R. (2026). Federated Self-Supervised Learning for Security-Oriented CCTV Camera-Hardware Analysis. Journal of Eye-Innovation in Security Analysis, 1(20261), 18. https://doi.org/10.68990/jiesa.202612
Vancouver
Ebrahim N, Nair R. Federated Self-Supervised Learning for Security-Oriented CCTV Camera-Hardware Analysis. Journal of Eye-Innovation in Security Analysis. 2026;1(20261):18. doi:10.68990/jiesa.202612
BibTeX
@article{Ebrahim2026,
author = {Ebrahim, Nadhem and Nair, Rajit},
title = {Federated Self-Supervised Learning for Security-Oriented CCTV Camera-Hardware Analysis},
journal = {Journal of Eye-Innovation in Security Analysis},
year = {2026},
volume = {1},
number = {20261},
pages = {18},
doi = {10.68990/jiesa.202612}
}
EUVIMED