WLD stands for WildLife Documentary, which is a dataset for animal target detection. The dataset contains 15 documentaries downloaded from YouTube, ranging from 9 minutes to 50 minutes in length, and a total of more than 747,000 frames. The dataset annotates the trajectories of more than 4,000 objects in 65 categories. One feature of this dataset is that all videos are accompanied by subtitles automatically generated by YouTube voice.

Citation

@inproceedings{chen2017discover,
author = {Kai Chen, Hang Song, Chen Change Loy, Dahua Lin},
title = {Discover and Learn New Objects from Documentaries},
booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = July,
year = {2017}
}