Dataset Viewer
The dataset viewer is not available for this subset.
Cannot get the split names for the config 'default' of the dataset.
Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
                  raise ValueError(
                  ...<2 lines>...
                  )
              ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
              
              The above exception was the direct cause of the following exception:
              
              Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/config/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

(We are now uploading the dataset files, because of the dataset size, it needs some time, in progress...) SGData is the semantic-aware dataset introduced with SGDFD. Real images come from all 1,000 ILSVRC2012 semantic categories; their fake counterparts are synthesised one-to-one from text prompts (derived automatically with QWEN3-VL) by seven state-of-the-art text-to-image generators. An additional semantic category — human faces — is sourced from FaceCaptionHQ-4M (images) and CelebV-Text (videos), and three video synthesis models provide unseen video forgeries. Faces in the SGData test set are deliberately left unprocessed (no alignment, no centre cropping), approximating realistic deployment.

Name Data Source Training Set Size Validation Set Size Test Set Size Total Size Num. Semantic Categories Num. Generation Models
SGData ILSVRC2012; FaceCaptionHQ-4M; CelebV-Text Real Images: 700,700
Fake Images: 700,700
Fake Videos: 0
Real Images: 70,070
Fake Images: 70,070
Fake Videos: 0
Real Images: 630,630
Fake Images: 630,630
Real Videos: 600
Fake Videos: 600
Real Images: 1,401,400
Fake Images: 1,401,400
Real Videos: 600
Fake Videos: 600
1,001 10

Generation models. Images (7): StableDiffusion 3.5, FLUX.2, PixelArt-Σ, Kandinsky 5.0, HiDream-I1, Lumina-Image 2.0, Qwen-Image — each producing 200 images for each of the 1,000 classes (7 × 200 × 1,000 = 1,400,000) plus 200 human-face images each. Videos (3): SANA-Video, HunyuanVideo 1.5, SkyReels-V2 — 200 human-face videos each, conditioned on CelebV-Text.

Layout of this repo:

SGData/
├── real/1000_categories/real.tar [only for real images sampled from ImageNet2012]
├── fake/<semantic_category>/<generators>.tar 
├── [real_videos/ you have to agree to the "Terms of access" of the CelebV-Text dataset to download the real videos by yourself]
└── fake_videos/<generators>.tar

The real images (FaceCaptionHQ-4M) and real videos (CelebV-Text), you have to download them by yourself because of "Terms of access". For ImageNet2012, we sampled 200 images from each of ImageNet2012 1000 classes and stored all theses sampled real images in this repo.

Downloads last month
79