DiDeMo / README.md
ziyjiang's picture
Update README.md
b24b6ce verified
metadata
configs:
  - config_name: default
    data_files:
      - split: train
        path: didemo_train.json
      - split: test
        path: didemo_test.json
task_categories:
  - text-to-video
  - text-retrieval
  - video-classification
language:
  - en
size_categories:
  - 10K<n<100K

Clone from friedrichor/DiDeMo.

About

DiDeMo contains 10K long-form videos from Flickr. For each video, ~4 short sentences are annotated in temporal order. We follow the existing works to concatenate those short sentences and evaluate ‘paragraph-to-video’ retrieval on this benchmark.

We adopt the official split:

  • Train: 8,395 videos, 8,395 captions (concatenate from 33,005 short captions)
  • Val: 1,065 videos, 1,065 captions (concatenate from 4,290 short captions) (We don't have the collection yet.)
  • Test: 1,004 videos, 1,004 captions (concatenate from 4,021 short captions)

Get Raw Videos

cat DiDeMo_Videos_mp4_train.tar.part-* | tar -vxf -
tar -xvf DiDeMo_Videos_mp4_test.tar

Official Release

Video Release: DiDeMoRelease


🌟 Citation

@inproceedings{anne2017didemo,
  title={Localizing moments in video with natural language},
  author={Anne Hendricks, Lisa and Wang, Oliver and Shechtman, Eli and Sivic, Josef and Darrell, Trevor and Russell, Bryan},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
  year={2017}
}