Research map / Image generation / Autoregressive and latent generation
Revisiting Deepfake Detection: BCNet for Robust Generalization Beyond Semantic Dependence
Jian Yang, Shibo Yao, Renshuai Tao, Chuangchuang Tan, Yao Zhao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add attention-guided semantic erasure (ASE) to baseline (ablation) | DINOv3 baseline · WildRF | accuracy: 90.1 -> 95.2 | Improved |
| add ASE to baseline (ablation) | DINOv3 baseline · Chameleon | accuracy: 76.3 -> 79.5 | Improved |
| add NPE to baseline (ablation) | DINOv3 baseline · Chameleon | accuracy: 76.3 -> 79.9 | Improved |
| add ASE and NPE together to baseline (ablation) | DINOv3 baseline · AIGI-Bench | accuracy: 84.2 -> 88.6 | Improved |
| add ASE and NPE together to baseline (ablation) | DINOv3 baseline · WildRF | accuracy: 90.1 -> 96.7 | Improved |
| add ASE and NPE together to baseline (ablation) | DINOv3 baseline · OpenSDI | accuracy: 70.5 -> 74.0 | Improved |
| apply ASE and NPE to real images as well as fake (ablation) | BCNet · 5 benchmarks (average) | accuracy: 87.7 -> 82.9 | Hurt |
| apply NPE only to real images as well (ablation) | BCNet · 5 benchmarks (average) | accuracy: 87.7 -> 82.8 | Hurt |
| apply ASE to real images as well (ablation) | BCNet · 5 benchmarks (average) | accuracy: 87.7 -> 86.1 | Hurt |
| use PGD instead of default normalized-gradient perturbation (ablation) | BCNet · 5 benchmarks (average) | accuracy: 87.7 -> 86.8 | No clear effect |
| use MI-FGSM instead of default perturbation (ablation) | BCNet · 5 benchmarks (average) | accuracy: 87.7 -> 85.9 | No clear effect |
| raise correction loss weights to 0.45:0.45 with CE 0.1 (vs 0.05:0.05:0.9) (ablation) | BCNet · 5 benchmarks (average) | accuracy: 84.6 -> 87.7 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- FerretNet — FerretNet: Efficient Synthetic Image Detection via Local Pixel Dependencies (NeurIPS 2025) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Standing on the Shoulders of Giants: Reprogramming Visual-Language Model for General Deepfake Detection (AAAI 2025 · similarity 0.89)
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (CVPR 2025 · similarity 0.89)
- Unlocking the Capabilities of Large Vision-Language Models for Generalizable and Explainable Deepfake Detection (ICML 2025 · similarity 0.88)
- Veritas: Generalizable Deepfake Detection via Pattern-Aware Reasoning (ICLR 2026 · similarity 0.88)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · similarity 0.88)
- Exploring Unbiased Deepfake Detection via Token-Level Shuffling and Mixing (AAAI 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
