You need to agree to share your contact information to access this dataset

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

EffectWorld is released for approved research use. Access requests are manually reviewed by the dataset owner. Please provide accurate affiliation and intended-use information and accept the data-use terms below.

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

EffectWorld

EffectWorld is a controlled-access paired video dataset for object removal with object-associated environmental effects. Each sample contains an unedited source, an edited target, and a binary object mask aligned over time.

Released subsets

Subset Samples Resolution Contents
EffectWorld 11,048 1280 x 704 source, target, binary mask
EffectWorld-Hardcase 1,000 1280 x 704 source, target, binary mask

The release contains only EffectWorld data. It does not include third-party datasets, model checkpoints, training code, evaluation outputs, or private logs.

Files

  • data_unpacked/effectworld/: 11,048 training samples.
  • data_unpacked/hardcase1000/: 1,000 hardcase samples.
  • benchmark/effectworld_eval33/: paired EffectWorld evaluation set.
  • benchmark/manifests/effectworld_eval33.csv: EffectWorld-Eval33 manifest.
  • benchmark/manifests/wild70_protocol_no_media.csv: EffectWorld-Wild70 IDs, labels, and evaluation protocol; third-party videos are not redistributed.

Legacy TAR frame shards are retained under data/ for archival use.

Benchmarks

EffectWorld-Eval33 is released with paired source, target, and mask videos. Its exact evaluated manifest is provided for reproducibility.

The EffectWorld-Wild70 IDs, labels, and evaluation protocol are provided in benchmark/manifests/wild70_protocol_no_media.csv. DAVIS and third-party web videos are not redistributed; users must obtain them from their original sources.

Access

Downloading requires manual approval from the dataset owner. Access is granted for the research purpose stated in the request and remains subject to DATA_USE_TERMS.md. Approval is not an unrestricted redistribution license.

Limitations

The data are synthetically rendered and do not represent every real-world capture condition. Masks identify primary removal targets; associated effects can extend beyond the mask. Users should independently assess suitability, bias, and safety for their intended application.

Citation

Citation information will be updated after the associated paper receives its final bibliographic record. Until then, cite this dataset repository and its versioned commit.

Downloads last month
50