The large scene video text dataset for scene video text spotting (LSVTD)

Ground Truth

XML file for LSVTD dataset

2021-06-01 (v. 1)

Contact author

Baorui Zou

Hikvision Research Institute

(+86) 18826072052

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.


scene video text detection, scene video text tracking, scene video text recognition


Following the 'Text in Videos' challenge, our ground truth files will be provided as a single XML file per video with the same format. The only difference is that we categorize the language as 'alphanumeric'  and 'non-alphanumeric'. Please refer to this website for details. 

xml example


No comments on this dataset yet.

Add your comment

In order to comment on a dataset you need to be logged on
Register Now!


In order to rate this dataset you need to be logged on
Register Now!