Dataset Viewer
Duplicate
The dataset viewer is not available for this split.
Cannot load the dataset split (in streaming mode) to extract the first rows.
Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
celebdf__YouTube-real__00284: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00285: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00286: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00287: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00288: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00289: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id37_id3_0006: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id13_id10_0008: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id30_id3_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id37_id21_0000: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id20_id19_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-real__id13_0011: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id28_id21_0006: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id35_id16_0004: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id1_id0_0007: list<item: int64>
  child 0, item: int64
celebdf__Celeb-real__id51_0005: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id28_id20_0008: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id53_id57_0005: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id28_id35_0008: list<item: int64>
  child 0, item: int64
celebdf__Celeb-real
...
_Celeb-synthesis__id31_id16_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id56_id55_0009: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id16_id0_0011: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id3_id9_0001: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id53_id52_0003: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id16_id23_0011: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00275: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id39_id47_0004: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id2_id6_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id9_id23_0009: list<item: int64>
  child 0, item: int64
celebdf__Celeb-real__id48_0000: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id6_id1_0007: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00274: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id16_id6_0001: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id9_id20_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id16_id6_0011: list<item: int64>
  child 0, item: int64
celebdf__Celeb-real__id3_0002: list<item: int64>
  child 0, item: int64
celebdf__Celeb-synthesis__id25_id21_0005: list<item: int64>
  child 0, item: int64
celebdf__YouTube-real__00188: list<item: int64>
  child 0, item: int64
to
{'celebdf__Celeb-real__id0_0001': List(Value('int64')), 'celebdf__Celeb-real__id10_0001': List(Value('int64')), 'celebdf__Celeb-real__id10_0007': List(Value('int64')), 'celebdf__Celeb-real__id11_0006': List(Value('int64')), 'celebdf__Celeb-real__id11_0008': List(Value('int64')), 'celebdf__Celeb-real__id12_0000': List(Value('int64')), 'celebdf__Celeb-real__id12_0005': List(Value('int64')), 'celebdf__Celeb-real__id13_0004': List(Value('int64')), 'celebdf__Celeb-real__id13_0011': List(Value('int64')), 'celebdf__Celeb-real__id13_0013': List(Value('int64')), 'celebdf__Celeb-real__id16_0007': List(Value('int64')), 'celebdf__Celeb-real__id16_0010': List(Value('int64')), 'celebdf__Celeb-real__id16_0011': List(Value('int64')), 'celebdf__Celeb-real__id17_0000': List(Value('int64')), 'celebdf__Celeb-real__id17_0002': List(Value('int64')), 'celebdf__Celeb-real__id19_0005': List(Value('int64')), 'celebdf__Celeb-real__id19_0007': List(Value('int64')), 'celebdf__Celeb-real__id1_0001': List(Value('int64')), 'celebdf__Celeb-real__id1_0006': List(Value('int64')), 'celebdf__Celeb-real__id1_0007': List(Value('int64')), 'celebdf__Celeb-real__id1_0008': List(Value('int64')), 'celebdf__Celeb-real__id20_0008': List(Value('int64')), 'celebdf__Celeb-real__id21_0006': List(Value('int64')), 'celebdf__Celeb-real__id21_0009': List(Value('int64')), 'celebdf__Celeb-real__id22_0000': List(Value('int64')), 'celebdf__Celeb-real__id22_0009': List(Value('int64')), 'celebdf__Celeb-real__id23_0001': List(Value('in
...
ouTube-real__00256': List(Value('int64')), 'celebdf__YouTube-real__00257': List(Value('int64')), 'celebdf__YouTube-real__00258': List(Value('int64')), 'celebdf__YouTube-real__00259': List(Value('int64')), 'celebdf__YouTube-real__00260': List(Value('int64')), 'celebdf__YouTube-real__00261': List(Value('int64')), 'celebdf__YouTube-real__00262': List(Value('int64')), 'celebdf__YouTube-real__00263': List(Value('int64')), 'celebdf__YouTube-real__00264': List(Value('int64')), 'celebdf__YouTube-real__00265': List(Value('int64')), 'celebdf__YouTube-real__00266': List(Value('int64')), 'celebdf__YouTube-real__00267': List(Value('int64')), 'celebdf__YouTube-real__00268': List(Value('int64')), 'celebdf__YouTube-real__00269': List(Value('int64')), 'celebdf__YouTube-real__00270': List(Value('int64')), 'celebdf__YouTube-real__00271': List(Value('int64')), 'celebdf__YouTube-real__00272': List(Value('int64')), 'celebdf__YouTube-real__00273': List(Value('int64')), 'celebdf__YouTube-real__00274': List(Value('int64')), 'celebdf__YouTube-real__00275': List(Value('int64')), 'celebdf__YouTube-real__00276': List(Value('int64')), 'celebdf__YouTube-real__00277': List(Value('int64')), 'celebdf__YouTube-real__00278': List(Value('int64')), 'celebdf__YouTube-real__00279': List(Value('int64')), 'celebdf__YouTube-real__00280': List(Value('int64')), 'celebdf__YouTube-real__00281': List(Value('int64')), 'celebdf__YouTube-real__00282': List(Value('int64')), 'celebdf__YouTube-real__00283': List(Value('int64'))}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 149, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 129, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 544, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 3052, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 558, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 361, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2381, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              celebdf__YouTube-real__00284: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00285: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00286: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00287: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00288: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00289: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id37_id3_0006: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id13_id10_0008: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id30_id3_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id37_id21_0000: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id20_id19_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-real__id13_0011: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id28_id21_0006: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id35_id16_0004: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id1_id0_0007: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-real__id51_0005: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id28_id20_0008: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id53_id57_0005: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id28_id35_0008: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-real
              ...
              _Celeb-synthesis__id31_id16_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id56_id55_0009: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id16_id0_0011: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id3_id9_0001: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id53_id52_0003: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id16_id23_0011: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00275: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id39_id47_0004: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id2_id6_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id9_id23_0009: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-real__id48_0000: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id6_id1_0007: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00274: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id16_id6_0001: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id9_id20_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id16_id6_0011: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-real__id3_0002: list<item: int64>
                child 0, item: int64
              celebdf__Celeb-synthesis__id25_id21_0005: list<item: int64>
                child 0, item: int64
              celebdf__YouTube-real__00188: list<item: int64>
                child 0, item: int64
              to
              {'celebdf__Celeb-real__id0_0001': List(Value('int64')), 'celebdf__Celeb-real__id10_0001': List(Value('int64')), 'celebdf__Celeb-real__id10_0007': List(Value('int64')), 'celebdf__Celeb-real__id11_0006': List(Value('int64')), 'celebdf__Celeb-real__id11_0008': List(Value('int64')), 'celebdf__Celeb-real__id12_0000': List(Value('int64')), 'celebdf__Celeb-real__id12_0005': List(Value('int64')), 'celebdf__Celeb-real__id13_0004': List(Value('int64')), 'celebdf__Celeb-real__id13_0011': List(Value('int64')), 'celebdf__Celeb-real__id13_0013': List(Value('int64')), 'celebdf__Celeb-real__id16_0007': List(Value('int64')), 'celebdf__Celeb-real__id16_0010': List(Value('int64')), 'celebdf__Celeb-real__id16_0011': List(Value('int64')), 'celebdf__Celeb-real__id17_0000': List(Value('int64')), 'celebdf__Celeb-real__id17_0002': List(Value('int64')), 'celebdf__Celeb-real__id19_0005': List(Value('int64')), 'celebdf__Celeb-real__id19_0007': List(Value('int64')), 'celebdf__Celeb-real__id1_0001': List(Value('int64')), 'celebdf__Celeb-real__id1_0006': List(Value('int64')), 'celebdf__Celeb-real__id1_0007': List(Value('int64')), 'celebdf__Celeb-real__id1_0008': List(Value('int64')), 'celebdf__Celeb-real__id20_0008': List(Value('int64')), 'celebdf__Celeb-real__id21_0006': List(Value('int64')), 'celebdf__Celeb-real__id21_0009': List(Value('int64')), 'celebdf__Celeb-real__id22_0000': List(Value('int64')), 'celebdf__Celeb-real__id22_0009': List(Value('int64')), 'celebdf__Celeb-real__id23_0001': List(Value('in
              ...
              ouTube-real__00256': List(Value('int64')), 'celebdf__YouTube-real__00257': List(Value('int64')), 'celebdf__YouTube-real__00258': List(Value('int64')), 'celebdf__YouTube-real__00259': List(Value('int64')), 'celebdf__YouTube-real__00260': List(Value('int64')), 'celebdf__YouTube-real__00261': List(Value('int64')), 'celebdf__YouTube-real__00262': List(Value('int64')), 'celebdf__YouTube-real__00263': List(Value('int64')), 'celebdf__YouTube-real__00264': List(Value('int64')), 'celebdf__YouTube-real__00265': List(Value('int64')), 'celebdf__YouTube-real__00266': List(Value('int64')), 'celebdf__YouTube-real__00267': List(Value('int64')), 'celebdf__YouTube-real__00268': List(Value('int64')), 'celebdf__YouTube-real__00269': List(Value('int64')), 'celebdf__YouTube-real__00270': List(Value('int64')), 'celebdf__YouTube-real__00271': List(Value('int64')), 'celebdf__YouTube-real__00272': List(Value('int64')), 'celebdf__YouTube-real__00273': List(Value('int64')), 'celebdf__YouTube-real__00274': List(Value('int64')), 'celebdf__YouTube-real__00275': List(Value('int64')), 'celebdf__YouTube-real__00276': List(Value('int64')), 'celebdf__YouTube-real__00277': List(Value('int64')), 'celebdf__YouTube-real__00278': List(Value('int64')), 'celebdf__YouTube-real__00279': List(Value('int64')), 'celebdf__YouTube-real__00280': List(Value('int64')), 'celebdf__YouTube-real__00281': List(Value('int64')), 'celebdf__YouTube-real__00282': List(Value('int64')), 'celebdf__YouTube-real__00283': List(Value('int64'))}
              because column names don't match

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.

DAVID-Net: Preprocessed Multimodal Forensic Shards & Foundation Features

Official dataset repository for DAVID-Net (Decoupled Audio-Visual Forensic Network), providing pre-extracted foundation tensor representations and preprocessed binary clip caches across seven benchmark media forensics corpora.

Paper GitHub Model Experiments Live Demo License: MIT


πŸ“Œ Dataset Overview & Scientific Purpose

Raw audio-visual deepfake benchmarks span hundreds of gigabytes of uncompressed video, requiring computationally intensive face tracking, facial landmark alignment, perioral mouth cropping, and acoustic resynthesis.

To ensure turnkey reproducibility and open science, this repository provides 809 curated tensor shards across two standardized operational tiers:

  1. clipcache/ (366 files: .bin + .json indexed pairs):
    • High-throughput binary tensor caches storing pre-decoded, YuNet face-aligned, perioral mouth-cropped ($224 \times 224$), and loudness-normalized clips.
    • Enables rapid PyTorch dataloading without raw video container decoding overhead.
  2. features/fs-a3b70d8acd/ (442 files: .safetensors + .json metadata pairs):
    • Pre-extracted foundation representation embeddings in SafeTensors format.
    • Visual Stream: Frozen VideoMAE ViT-Base spatio-temporal tokens ($128 \times 768$).
    • Acoustic Stream: Frozen WavLM Base+ speech tokens ($50 \times 768$).
    • Enables complete replication of the 160-run evaluation campaign in minutes on standard GPU hardware without backbone re-encoding.

πŸ“Š Benchmark Corpora Coverage & Shard Inventory

Benchmark Corpus Forensic Modality ClipCache Shards (.bin + .json) Feature Shards (.safetensors + .json) Forensic Role in Study
FakeAVCeleb Multimodal ($AV$) 169 shards 170 shards Primary 4-quadrant benchmark (RVRA, RVFA, FVRA, FVFA) under strict identity-disjoint partitions ($0.0%$ leakage).
DFDC-10 Multimodal ($AV$) 10 shards 10 shards In-the-wild facial manipulation under heavy compression artifacts (5,000 clips).
Celeb-DF v2 Video ($V$) 2 shards 2 shards High-quality visual deepfake benchmark testing spatial boundary transfer (518 clips).
DeepFakeTIMIT Video ($V$) 2 shards 2 shards Autoencoder face swap benchmark (640 clips, $99.8%$ detection rate).
ASVspoof 2019 LA Audio ($A$) β€” 18 shards Acoustic logical access anti-spoofing benchmark (25,380 evaluation utterances; $+19.1%$ AUC transfer gain).
In-the-Wild Audio Audio ($A$) β€” 9 shards Unconstrained social-media audio spoofing evaluation (31,779 utterances; $+12.2%$ AUC gain).
WaveFake Audio ($A$) β€” 3 shards Acoustic generalization across 11 unseen neural vocoder architectures (11,000 utterances).
QACP Proxy Pool Multimodal ($AV$) β€” 6 shards Self-supervised proxy contrastive pretraining data pool.

βš™οΈ Data Preprocessing & Feature Extraction Parameters

As documented in features/fs-a3b70d8acd/FEATURES_META.json:

  • Face Detection: OpenCV YuNet detector (box_scale=1.8, decode_max_side=640, detect_frames=4).
  • Spatial Resolution: $224 \times 224$ pixels, normalized using standard ImageNet statistics ($\mu = [0.485, 0.456, 0.406]$, $\sigma = [0.229, 0.224, 0.225]$).
  • Temporal Framing: 16 uniformly sampled frames spanning a 4.0-second window.
  • Acoustic Audio: 16,kHz mono, loudness-normalized raw speech waveforms.
  • Visual Token Dimension: $L_v = 128$ ($8$ temporal tubelets $\times 4 \times 4$ spatial grid), $d = 768$.
  • Acoustic Token Dimension: $L_a = 50$ (each token pooling 80,ms of speech), $d = 768$.

πŸ’» Loading Features in PyTorch

from huggingface_hub import hf_hub_download
from safetensors.torch import load_file
import json

repo_id = "MIHMahmudEli/david-net-av"

# 1. Download shard features (SafeTensors) and metadata
feat_path = hf_hub_download(
    repo_id=repo_id,
    filename="features/fs-a3b70d8acd/fakeavceleb/shard_00000.safetensors",
    repo_type="dataset"
)
meta_path = hf_hub_download(
    repo_id=repo_id,
    filename="features/fs-a3b70d8acd/fakeavceleb/shard_00000.json",
    repo_type="dataset"
)

# 2. Inspect tensors
tensors = load_file(feat_path)
with open(meta_path, "r") as f:
    meta = json.load(f)

print(f"Loaded {len(tensors)} tensors across {len(meta)} clip entries.")
# Keys: '{clip_id}_video' -> [128, 768], '{clip_id}_audio' -> [50, 768]

πŸ”’ Cryptographic Partitioning & Leakage Audit

All evaluation splits follow strict cryptographic verification:

  • Training, validation, and test manifests are partitioned by subject identity with 0.0% overlap.
  • Manifest hashes and split definitions are permanently version-controlled in SPLITS_SHA256.json.

βš–οΈ Ethics & Data Governance

All benchmark corpora included in this feature repository were acquired in accordance with their respective institutional research agreements and licenses. No uncompressed, raw facial video or identifiable biometric recordings are redistributed. Pre-extracted feature vectors and cropped patches are provided strictly for defensive forensic verification, academic reproducibility, and peer-review audits.


πŸ“œ Citation

@article{hossain2026davidnet,
  title={DAVID-Net: Disentangled Audio-Visual Forensic Attribution under Identity-Disjoint Evaluation},
  author={Hossain, Moshin Ibna and Khan, Alif Hasan and Shakib, Shadman and Jam, MD. Yeahyea and Hasan, Mohammad Mahmudul},
  journal={Machine Learning with Applications},
  year={2026},
  note={Under review}
}

@thesis{hossain2026thesis,
  title={Decoupled Multimodal Forensic Attribution and Desynchronization Detection},
  author={Hossain, Moshin Ibna and Khan, Alif Hasan and Shakib, Shadman and Jam, MD. Yeahyea},
  school={American International University-Bangladesh},
  year={2026}
}
Downloads last month
124

Space using MIHMahmudEli/david-net-av 1