Word-Anchored Temporal Forgery Localization

TL;DR

This paper introduces WAFL, a novel word-anchored approach for temporal forgery localization that improves accuracy and efficiency by shifting from continuous to discrete word-level classification, utilizing a forensic feature realignment module and a specialized loss function.

Contribution

The paper proposes a new paradigm for TFL based on word-level classification, a forensic feature realignment module, and an artifact-centric asymmetric loss, advancing the state-of-the-art in forgery localization.

Findings

WAFL outperforms existing methods in localization accuracy.

It requires fewer parameters and is computationally efficient.

Effective in both in- and cross-dataset evaluations.

Abstract

Current temporal forgery localization (TFL) approaches typically rely on temporal boundary regression or continuous frame-level anomaly detection paradigms to derive candidate forgery proposals. However, they suffer not only from feature granularity misalignment but also from costly computation. To address these issues, we propose word-anchored temporal forgery localization (WAFL), a novel paradigm that shifts the TFL task from temporal regression and continuous localization to discrete word-level binary classification. Specifically, we first analyze the essence of temporal forgeries and identify the minimum meaningful forgery units, word tokens, and then align data preprocessing with the natural linguistic boundaries of speech. To adapt powerful pre-trained foundation backbones for feature extraction, we introduce the forensic feature realignment (FFR) module, mapping representations from the pre-trained semantic space to a discriminative forensic manifold. This allows subsequent lightweight linear classifiers to efficiently perform binary classification and accomplish the TFL task. Furthermore, to overcome the extreme class imbalance inherent to forgery detection, we design the artifact-centric asymmetric (ACA) loss, which breaks the standard precision-recall trade-off by dynamically suppressing overwhelming authentic gradients while asymmetrically prioritizing subtle forensic artifacts. Extensive experiments demonstrate that WAFL significantly outperforms state-of-the-art approaches in localization performance under both in- and cross-dataset settings, while requiring substantially fewer learnable parameters and operating at high computational efficiency.