Hugging Face's logo

Datasets:
LikhithaGeddam
/
fakevideodetection-private

Tasks:
Video Classification
Modalities:
Video
Size:
10K - 100K
Libraries:
Datasets
License:
Dataset card Data Studio Files
xet
 Community
1

You need to agree to share your Hugging Face information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Deepfake Detection Dataset (Private)

Private repository — not for redistribution outside authorized use.

A combined video dataset for deepfake/synthetic-media detection, organized into train/, val/, and test/ splits, each containing real/ and fake/ subfolders.

Sources and licensing

CREMA-D (real)

FaceForensics++ (real + fake)

  • License: non-commercial research use, individually granted by Technical University of Munich (TUM) after agreeing to the FaceForensics Terms of Use.
  • Access obtained via the official request process (not a third-party mirror).
  • Not to be redistributed outside this private, authorized use.

Celeb-DF-v2 (real + fake)

  • License: non-commercial research use, access granted directly by the original authors after signing their usage agreement.
  • Not to be redistributed outside this private, authorized use.

AI-generated clips (fake)

  • Generated using Adobe Firefly, using the Kling 3.0 model (a third-party partner model made available through the Firefly platform).
  • Output ownership: retained by the generating user per Adobe's Generative AI terms.
  • Commercial-use / IP indemnification: Adobe's commercial-safe guarantee and indemnification are described for Adobe's own Firefly models; it is not fully confirmed whether the same indemnification applies identically to outputs from partner models (Kling) accessed via Firefly. Treat these clips as non-commercial-safe unless separately verified.
  • No real individuals' likenesses were used as generation input. (Update this line if that's not accurate — see note below.)

Overall usage

This dataset combines multiple non-commercial research sources. The most restrictive terms govern the whole dataset: non-commercial research use only, private access only, no redistribution.

Splits

Split Real Fake Total
Train 6,532 9,634 16,166
Val 1,400 1,904 3,304
Test 1,400 2,065 3,465

Structure

``` train/real/*.mp4 train/fake/*.mp4 val/real/*.mp4 val/fake/*.mp4 test/real/*.mp4 test/fake/*.mp4 metadata.jsonl ```

Citations

  • Cozzolino et al., "FaceForensics++: Learning to Detect Manipulated Facial Images," ICCV 2019.
  • Li et al., "Celeb-DF: A Large-scale Challenging Dataset for DeepFake Forensics," CVPR 2020.
  • Cao et al., "CREMA-D: Crowd-sourced Emotional Multimodal Actors Dataset," IEEE Trans. Affective Computing, 2014.
Downloads last month
241
Total file size:
44.8 GB