# Local Manipulation Video (LMV)

[Back to benchmarks](README.md) · [Back to home](../README.md)

LMV benchmarks preserve an authentic carrier and remain the historical backbone of video deepfake evaluation. They are especially useful for compression robustness, localized forensic residue, codec variation, and cross-dataset transfer.

| Date | Benchmark / Dataset | Paper | Focus | Venue |
| --- | --- | --- | --- | --- |
| 02/2026 | FAQ / Beyond Static Artifacts | [Beyond Static Artifacts: A Forensic Benchmark for Video Deepfake Reasoning in Vision Language Models](https://doi.org/10.48550/ARXIV.2602.21779)<br>Gu et al. | Forensic benchmark and instruction-tuning resource for temporal video deepfake reasoning in VLMs. | CVPR |
| 11/2025 | ExDDV | [Exddv: A new dataset for explainable deepfake detection in video](https://arxiv.org/abs/2503.14421)<br>Hondru et al. | Benchmark for explainable AIGC-V video detection. | arXiv |
| 09/2024 | DD-VQA | [Common sense reasoning for deepfake detection](https://doi.org/10.1007/978-3-031-73223-2_22)<br>Zhang et al. | AIGC-V detection VQA (image--question--answer triplets). | ECCV |
| 06/2024 | AI-Face | [Ai-face: A million-scale demographically annotated ai-generated face dataset and fairness benchmark](https://doi.org/10.1109/cvpr52734.2025.00332)<br>Lin et al. | Million-scale, demographically annotated AI-face dataset. | CVPR |
| 07/2023 | DeepfakeBench | [DeepfakeBench: A Comprehensive Benchmark of Deepfake Detection](https://proceedings.neurips.cc/paper_files/paper/2023/file/0e735e4b4f07de483cbe250130992726-Paper-Datasets_and_Benchmarks.pdf)<br>Yan et al. | Comprehensive AIGC-V detection benchmark. | NeurIPS |
| 06/2023 | DF-Platter | [DF-Platter: Multi-Face Heterogeneous Deepfake Dataset](https://doi.org/10.1109/cvpr52729.2023.00939)<br>Narayan et al. | Multi-face heterogeneous deepfake dataset. | CVPR |
| 01/2023 | CDDB | [A Continual Deepfake Detection Benchmark: Dataset, Methods, and Essentials](https://doi.org/10.1109/wacv56688.2023.00139)<br>Li et al. | Benchmark for easy, hard, and long-sequence AIGC-V detection. | WACV |
| 10/2021 | KoDF | [Kodf: A large-scale korean deepfake detection dataset](https://arxiv.org/abs/2103.10094)<br>Kwon et al. | Large-scale Korean AIGC-V detection dataset. | ICCV |
| 06/2021 | ForgeryNet | [ForgeryNet: A Versatile Benchmark for Comprehensive Forgery Analysis](https://doi.org/10.1109/cvpr46437.2021.00434)<br>He et al. | Mega-scale benchmark for image and video face-forgery analysis. | CVPR |
| 10/2020 | WildDeepfake | [WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection](https://doi.org/10.1145/3394171.3413769)<br>Zi et al. | In-the-wild dataset for AIGC-V detection. | ACM MM |
| 06/2020 | DFDC | [The DeepFake Detection Challenge (DFDC) Dataset](https://arxiv.org/abs/2006.07397)<br>Dolhansky et al. | Large-scale face-swapping dataset (mostly local forgeries). | arXiv |
| 05/2020 | DeeperForensics-1.0 | [DeeperForensics-1.0: A Large-Scale Dataset for Real-World Face Forgery Detection](https://doi.org/10.1109/cvpr42600.2020.00296)<br>Jiang et al. | Large-scale dataset for real-world face-forgery detection. | CVPR |
| 09/2019 | Celeb-DF | [Celeb-df: A large-scale challenging dataset for deepfake forensics](https://doi.org/10.1109/cvpr42600.2020.00327)<br>Li et al. | Large-scale challenging deepfake-forensics dataset. | CVPR |
| 01/2019 | FaceForensics++ | [Faceforensics++: Learning to detect manipulated facial images](https://doi.org/10.1109/iccv.2019.00009)<br>Rossler et al. | Forensics dataset with 1,000 original videos. | ICCV |