Status: Under Publication — Accepted at ICACA-2026, awaiting final publication.

Proposes a deepfake video detection framework that focuses processing on face regions combined with strategic frame sampling, improving both detection accuracy and computational efficiency over naive full-frame approaches.