wqz995's picture
Citation request wording
f3d4af5 verified
|
Raw History Blame Contribute Delete
3.75 kB
metadata
pretty_name: SurFITR-Localisation
license: cc-by-nc-4.0
task_categories:
  - image-segmentation
tags:
  - image-forgery-localisation
  - surveillance
  - multimedia-forensics
  - diffusion-editing
size_categories:
  - 100K<n<1M
extra_gated_prompt: >-
  SurFITR-Localisation contains pixel-level tampering masks for the SurFITR
  dataset. Access is granted for non-commercial research use under CC BY-NC 4.0.
  Please tell us who you are and how you intend to use the masks.
extra_gated_fields:
  Full name: text
  Affiliation: text
  Intended use: text
  I agree to use the data for non-commercial research only and not to redistribute it: checkbox
configs:
  - config_name: base
    data_files:
      - split: test
        path: data/base_preview.parquet
    features:
      - name: original
        dtype: image
      - name: tampered
        dtype: image
      - name: tampering_mask
        dtype: image
      - name: source_corpora
        dtype: string
      - name: type
        dtype: string
  - config_name: transfer
    data_files:
      - split: eval
        path: data/transfer_preview.parquet
    features:
      - name: original
        dtype: image
      - name: tampered
        dtype: image
      - name: tampering_mask
        dtype: image
      - name: source_corpora
        dtype: string
      - name: type
        dtype: string
      - name: model
        dtype: string

SurFITR-Localisation

Pixel-level tampering masks for the Surveillance Forgery Image Test Range (SurFITR), published at ACM Multimedia 2026. This repository holds only the masks. The images and labels are in the open companion repository wqz995/SurFITR-Detection; download both to work on localisation.

Files

File Size Contents
base_masks.tar.gz 214 MB 108,171 masks for SurFITR-Base (train and test)
transfer_masks.tar.gz 61 MB 29,985 masks for SurFITR-Transfer (train = Eval 1, test = Eval 2)
data/*_preview.parquet Viewer previews only

Each mask has the same resolution as its image and is stored as JPEG; white marks the tampered region. Real images have no mask.

Download

pip install -U huggingface_hub
hf auth login   # after your access request is approved

# images and labels (open)
hf download wqz995/SurFITR-Detection --repo-type dataset --local-dir SurFITR
mkdir -p SurFITR/base SurFITR/transfer
tar -xzf SurFITR/base.tar.gz -C SurFITR/base
tar -xzf SurFITR/transfer.tar.gz -C SurFITR/transfer

# masks (this repository)
hf download wqz995/SurFITR-Localisation --repo-type dataset --local-dir SurFITR
tar -xzf SurFITR/base_masks.tar.gz -C SurFITR/base
tar -xzf SurFITR/transfer_masks.tar.gz -C SurFITR/transfer

After extraction every tampered image fake/<corpus>/<split>/NNNNNNNN.jpg has its mask at fake/<corpus>/<split>/NNNNNNNN_mask.jpg.

License

CC BY-NC 4.0, for research use only.

Citation

If you find SurFITR useful for your research, please cite our paper.

@inproceedings{wang2026surfitr,
  author    = {Wang, Qizhou and Pang, Guansong and Leckie, Christopher},
  title     = {SurFITR: A Dataset for Surveillance Image Forgery Detection and Localisation},
  booktitle = {Proceedings of the 34th ACM International Conference on Multimedia},
  series    = {MM '26},
  year      = {2026},
  pages     = {16972--16979},
  doi       = {10.1145/3767308.3838668},
  url       = {https://doi.org/10.1145/3767308.3838668}
}