Research map / Detection and segmentation / Open-set and few-shot detection
OpenForensics: Large-Scale Challenging Dataset for Multi-Face Forgery Detection and Segmentation In-the-Wild
Trung-Nghia Le, Huy H. Nguyen, Junichi Yamagishi, Isao Echizen
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 |
|---|---|---|---|
| use OpenForensics instead of Celeb-DF as the source of forged faces | human viewers · image realism user study | mean opinion score: 4.0 vs 3.2 | Improved |
| use OpenForensics instead of Celeb-DF as the source of forged faces | human viewers · face forgery classification | human accuracy: 59.7% vs 68.7% | Hurt |
| use OpenForensics instead of FaceForensics++ | human viewers · face forgery classification | human accuracy: 59.7% vs 82.0% | Hurt |
| increase the number of forged faces in an image | human participants · multi-face forgery detection | human detection accuracy: 86% -> 0% | Hurt |
| use BlendMask instead of MaskRCNN | detector with FPN-ResNet50 · OpenForensics detection (test-dev) | AP: 79.2 -> 87.0 | Improved |
| use BlendMask instead of YOLACT | detector with FPN-ResNet50 · OpenForensics detection (test-dev) | AP: 68.1 -> 87.0 | Improved |
| evaluate on unseen test-challenge images instead of test-dev | BlendMask · OpenForensics detection | AP: 87.0 -> 53.9 | Hurt |
| evaluate on unseen test-challenge images instead of test-dev | MaskRCNN · OpenForensics detection | AP: 79.2 -> 42.1 | Hurt |
| use YOLACT++ instead of MaskRCNN on unseen images | detector with FPN-ResNet50 · OpenForensics detection (test-challenge) | AP: 42.1 -> 53.7 | Improved |
| use CenterMask on unseen images | CenterMask · OpenForensics detection (test-challenge) | AP: 85.5 -> 0.03 | Hurt |
| use BlendMask for segmentation instead of detection-only boxes | BlendMask · OpenForensics segmentation (test-dev) | AP: 89.2 | Improved |
| use single-stage dense detectors instead of two-stage sparse ones | forgery detectors · OpenForensics detection/segmentation | oLRP FP and FN | Mixed |
Closest papers by meaning
- DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection (CVPR 2020 · similarity 0.91)
- DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · similarity 0.89)
- FaceForensics++: Learning to Detect Manipulated Facial Images (ICCV 2019 · similarity 0.89)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.88)
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.88)
- A Hitchhiker's Guide to Fine-Grained Face Forgery Detection Using Common Sense Reasoning (NeurIPS 2024 · similarity 0.87)
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.
