Datasets:

friedrichor
/

MSR-VTT

Tasks:

Video Classification

Modalities:

Formats:

Languages:

Size:

Libraries:

Dataset card Data Studio Files Files and versions Community

friedrichor commited on 4 days ago

Commit

81f238b

·

verified ·

1 Parent(s): fabb858

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -28,7 +28,7 @@ size_categories:
 We adopt the standard `1K-A split` protocol, which was introduced in [JSFusion](https://openaccess.thecvf.com/content_ECCV_2018/html/Youngjae_Yu_A_Joint_Sequence_ECCV_2018_paper.html) and has since become the de facto benchmark split in the `Text-Video Retrieval` field.
 Train:
-- train_7k: 7,100 videos, 140,200 captions
 - train_9k: 9,000 videos, 180,000 captions
 Test:

 We adopt the standard `1K-A split` protocol, which was introduced in [JSFusion](https://openaccess.thecvf.com/content_ECCV_2018/html/Youngjae_Yu_A_Joint_Sequence_ECCV_2018_paper.html) and has since become the de facto benchmark split in the `Text-Video Retrieval` field.
 Train:
+- train_7k: 7,010 videos, 140,200 captions
 - train_9k: 9,000 videos, 180,000 captions
 Test: