# Segmentation-Guided Spatial Indexing for Generalizable and Explainable Deepfake Detection

arXiv: https://arxiv.org/abs/2606.00098 · HTML: https://mlsanity.com/paper/2606.00098/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 9 of its 13 references that are in the map.
- First posted on arXiv 25 May 2026.

## Builds on (most cited first, 9 of 9)

- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [FaceForensics++: Learning to Detect Manipulated Facial Images](https://mlsanity.com/paper/1901.08971/) (2019)
- [Celeb-DF: A Large-scale Challenging Dataset for DeepFake Forensics](https://mlsanity.com/paper/1909.12962/) (2019)
- [Towards Universal Fake Image Detectors that Generalize Across Generative Models](https://mlsanity.com/paper/2302.10174/) (2023)
- [Multi-attentional Deepfake Detection](https://mlsanity.com/paper/2103.02406/) (2021)
- [Lips Don't Lie: A Generalisable and Robust Approach to Face Forgery Detection](https://mlsanity.com/paper/2012.07657/) (2020)
- [General Facial Representation Learning in a Visual-Linguistic Manner](https://mlsanity.com/paper/2112.03109/) (2021)
- [Unlocking the Hidden Potential of CLIP in Generalizable Deepfake Detection](https://mlsanity.com/paper/2503.19683/) (2025)
- [Face2Parts: Exploring Coarse-to-Fine Inter-Regional Facial Dependencies for Generalized Deepfake Detection](https://mlsanity.com/paper/2603.26036/) (2026)

## Abstract (arXiv)

We introduce segmentation-guided spatial indexing for generalizable and explainable deepfake detection. The key idea reverses the standard design order: rather than pooling all facial tokens and classifying afterward, we first select semantically meaningful patch tokens, then pool only those. A frozen FaRL parser assigns each DINOv3 ViT-L/16 patch token a semantic label; non-target tokens are discarded; a linear probe classifies the retained region. This spatial indexing exploits DINOv3's patch-level spatial consistency, the same property that enables emergent segmentation, to present the probe with a purer regional subspace where manipulation-relevant evidence is less diluted by whole-face cues. Region attribution is structural: when the mouth model predicts fake, the decision used only mouth tokens, not an overlaid saliency map. On Celeb-DF v2, the mouth-indexed probe achieves AUC 0.905, outperforming LipForensics (+8.1 pp) and Xception (+16.9 pp), with no DINOv3 or FaRL fine-tuning and no target-domain data. Ablations isolate the mechanism: replacing regional selection with DINOv3's CLS token drops Celeb-DF v2 AUC by 26.4 pp; replacing DINOv3 with FaRL features drops it by 20.9 pp. Both DINOv3 representation and the spatial index are independently necessary; neither alone approaches the full system.