| --- |
| license: mit |
| --- |
| # TalkingHeadBench |
|
|
| ## Overview |
|
|
| The **TalkingHeadBench(THB)** is a curated dataset designed to support the training and evaluation of deepfake detection models, especially in audio-visual and cross-method generalization scenarios. It includes synthetic videos generated using six modern face animation techniques: |
|
|
| - **LivePortrait** |
| - **AniPortraitAudio** |
| - **AniPortraitVideo** |
| - **Hallo** |
| - **Hallo2** |
| - **EmoPortrait** |
|
|
| Each video is named using the format: |
|
|
| [image]--[driving_signals]--[generation_method].mp4 |
| |
| - `image`: identity image from **FFHQ** |
| - `driving_signals`: facial motion and optionally audio from **CelebV-HQ** |
| - `generation_method`: the name of the generator used |
|
|
| ## Dataset Structure |
|
|
| ``` |
| TalkingHeadBench/ |
| ├── fake/ |
| │ ├── [generator_name]/[split]/*.mp4 |
| │ ├── additional_dataset/[generator_name]/*.mp4 # Additional evaluation-only dataset generated using MAGI-1 and Hallo3. |
| ├── audio/ |
| │ ├── fake/*.wav # From CelebV-HQ driving signals |
| │ ├── ff++/*.wav # From FaceForensics++ YouTube videos |
| ├── true/ |
| │ ├── true_dataset_split.json # Filenames of true (non-deepfake) videos collected from CelebV-HQ and FaceForensics++ |
| |
| ``` |
|
|
| - Each generator has three splits: `train`, `val`, and `test` |
| - Training and testing sets come from **disjoint identity pools** |
| - ~300 fake videos per generator are used for training |
| - 50 videos per generator are held out as validation |
| - Testing uses entirely unseen identities |
|
|
| ## True Dataset |
| - For training and evaluating purposes, we added true (non-deepfake) videos to the process at an approximately 1:1 ratio |
| - We used CelebV-HQ and FaceForensics++ (`faceforensics++/original_sequences/youtube/raw/videos`) for our source of true videos |
| - All the true videos are checked against both driving signals and images to ensure no id leakage. |
|
|
| ## Audio Details |
|
|
| - **Fake Audio** (`audio/fake/`): Extracted from the CelebV-HQ videos that drove the generation. |
| - **True Audio** (`audio/ff++/`): Original audio from the FaceForensics++ YouTube videos, named as `xxx.wav` (three-digit index). |
| - Maps directly to: `faceforensics++/original_sequences/youtube/raw/videos` |
| - We provide 704 audio clips out of 1000, due to public unavailability of some videos. |
|
|
| ## Applications |
|
|
| - Audio-visual deepfake detection |
| - Modality-specific detection (audio-only or video-only) |
| - Cross-generator generalization testing |
| - Audio-video consistency evaluation |
|
|
| ## Licensing and Attribution |
|
|
| Please ensure compliance with the original licenses: |
|
|
| - [FFHQ License](https://github.com/NVlabs/ffhq-dataset#license) |
| - [CelebV-HQ License](https://github.com/CelebV-HQ/CelebV-HQ#license) |
| - [FaceForensics++ License](https://github.com/ondyari/FaceForensics#license) |
|
|
| ## Citation |
|
|
| If you use this dataset in your research, please cite the relevant original sources (FFHQ, CelebV-HQ, FaceForensics++) and this repository. |
|
|
