Description
The V3C2 dataset (drawn from a larger V3C video dataset) is composed of 9760 Vimeo videos (1.6 TB, 1300 h) with Creative Commons licenses and mean duration of 8 min. All videos will have some metadata available e.g., title, keywords, and description in json files. The dataset has been segmented into 1,425,454 short video segments according to the provided master shot boundary files. In addition, Keyframes and thumbnails per video segment have been extracted and available. For more details, please consult the TRECVID website: trecvid.nist.gov
Resources
| Name | Format | Description | Link |
|---|---|---|---|
| Queries and Ground truth | HTML | Please use this URL to access the 2019, 2020, and 2021 queries and ground truth files used with the V3C1 dataset. | Open resource |
| Data Agreement | Text | Please fill, sign, and submit this data agreement and once processed, you will be notified with the access information to download the dataset. | Open resource |
| The Ad-hoc Video Search Task guidelines | Unknown | This is the AVS (Ad-hoc Video Search) Task guidelines page. The guidelines highlight the task definition, run formats, conditions, and metrics. | Open resource |
| V3C1 master shot reference | Text | Readme for the master shot reference:http://www-nlpir.nist.gov/projects/tv2019/pastdata/ad.hoc.search/V3C1.shotNames.readme | Open resource |