aohan12138 / Lstn_fdst_dataset

We collected 100 videos captured from 13 different scenes, and FDST dataset contains 150,000 frames, with a total of 394,081 annotated heads.

Lstn_fdst_dataset

We collected 100 videos captured from 13 different scenes, and FDST dataset contains 150,000 frames, with a total of 394,081 annotated heads, in particular,the training set of FDST dataset consists of 60 videos, 9000 frames and the testing set contains the remaining 40 videos, 6000 frames.

Download FDST datasets

Please manually download our dataset: [Baidu] (password:sgt1) or [GoogleDrive].

Citation

If you find this useful, please cite our work as follows:

@article{fang2019locality,
  title={Locality-constrained Spatial Transformer Network for Video Crowd Counting},
  author={Fang, Yanyan and Zhan, Biyun and Cai, Wandi and Gao, Shenghua and Hu, Bo},
  journal={arXiv preprint arXiv:1907.07911},
  year={2019}
}