AIFIND: Artifact-Aware Interpreting Fine-Grained Alignment for Incremental Face Forgery Detection

2026-04-17

AI Summary

This work addresses feature drift and catastrophic forgetting in incremental face forgery detection, which arise from the lack of explicit feature constraints. To mitigate these issues, the authors propose a forgery-artifact-aware fine-grained alignment framework that, for the first time, leverages low-level forgery artifacts as high-level semantic anchors to construct a stable semantic coordinate system. Alignment of visual features to these invariant semantic anchors is achieved through an artifact-driven semantic prior generator and an artifact-probe attention mechanism. Additionally, an adaptive decision coordinator combined with an angular relation preservation strategy maintains geometric consistency among classifiers. Without requiring data replay, the proposed method significantly outperforms existing approaches across multiple incremental learning protocols, effectively alleviating forgetting and enhancing cross-forgery-type detection performance.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

As forgery types continue to emerge consistently, Incremental Face Forgery Detection (IFFD) has become a crucial paradigm. However, existing methods typically rely on data replay or coarse binary supervision, which fails to explicitly constrain the feature space, leading to severe feature drift and catastrophic forgetting. To address this, we propose AIFIND, Artifact-Aware Interpreting Fine-Grained Alignment for Incremental Face Forgery Detection, which leverages semantic anchors to stabilize incremental learning. We design the Artifact-Driven Semantic Prior Generator to instantiate invariant semantic anchors, establishing a fixed coordinate system from low-level artifact cues. These anchors are injected into the image encoder via Artifact-Probe Attention, which explicitly constrains volatile visual features to align with stable semantic anchors. Adaptive Decision Harmonizer harmonizes the classifiers by preserving angular relationships of semantic anchors, maintaining geometric consistency across tasks. Extensive experiments on multiple incremental protocols validate the superiority of AIFIND.

Problem

• Incremental Face Forgery Detection
• feature drift
• catastrophic forgetting
• semantic anchors
• fine-grained alignment

Innovation

• Incremental Learning
• Face Forgery Detection
• Semantic Anchors
• Artifact-Aware Alignment
• Catastrophic Forgetting