Research map / Vision-language and multimodal / Multimodal LLMs
Common Sense Reasoning for Deep Fake Detection
Yue Zhang, Ben Colman, Xiao Guo, Ali Shahriyari, Gaurav Bharaj
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 text contrastive loss to BLIP fine-tuning on DD-VQA (ablation) | BLIP · DD-VQA test | detection accuracy: 0.8168 -> 0.8365 | Improved |
| add image contrastive loss to BLIP fine-tuning on DD-VQA (ablation) | BLIP · DD-VQA test | detection accuracy: 0.8168 -> 0.8487 | Improved |
| add both text and image contrastive losses (ablation) | BLIP · DD-VQA test | detection F1: 0.8642 -> 0.9007 | Improved |
| add recall consideration: contrastive losses reduce recall (ablation) | BLIP · DD-VQA test | detection recall: 0.9596 -> 0.9341 | Hurt |
| integrate DD-VQA representations (BLIP-TI) into XceptionNet | XceptionNet · FF++ (c40) intra-testing | AUC: 89.30 -> 92.24 | Improved |
| integrate DD-VQA representations into HiFi-Net | HiFi-Net · WildDeepfake cross-testing | AUC: 65.22 -> 70.03 | Improved |
| integrate DD-VQA representations into SBIs | SBIs · Celeb-DF cross-testing | AUC: 93.18 -> 93.98 | Improved |
| integrate DD-VQA representations into RECCE | RECCE · WildDeepfake cross-testing | AUC: 64.31 -> 69.46 | Improved |
| use plain pretrained BLIP weights in RECCE instead of no BLIP (ablation) | RECCE · WildDeepfake cross-testing | AUC: 64.31 -> 67.24 | Improved |
| use plain pretrained BLIP weights in RECCE (ablation) | RECCE · FF++ (c40) intra-testing | ACC: 91.03 -> 89.22 | Hurt |
| add textual explanations to answers (ablation) | BLIP on DD-VQA · DD-VQA test | detection accuracy: 0.7978 -> 0.8168 | Improved |
| add textual explanations to answers, cross-tested (ablation) | BLIP · SeqDeepfake cross-testing | detection accuracy: 0.5216 -> 0.5648 | Improved |
Closest papers by meaning
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.90)
- A Hitchhiker's Guide to Fine-Grained Face Forgery Detection Using Common Sense Reasoning (NeurIPS 2024 · similarity 0.90)
- Veritas: Generalizable Deepfake Detection via Pattern-Aware Reasoning (ICLR 2026 · similarity 0.88)
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (CVPR 2025 · similarity 0.88)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.88)
- Exposing the Deception: Uncovering More Forgery Clues for Deepfake Detection (AAAI 2024 · 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.
