This is the Beagle dog dataset used in the paper "Three-dimensional surface motion capture of multiple freely moving pigs using MAMMAL" (Nature Communications 2023). If you have any problem, contact Liang An (anliang at mail dot tsinghua dot edu dot cn).
You could access the dataset by two ways:
Google Drive: https://drive.google.com/drive/folders/1bRYKzB1-ClJwHGsk1g6Opi04eijASp0U?usp=sharing.
Baidu Pan: https://pan.baidu.com/s/1ZxPvXKjECu-HKLx3OSg9eg?pwd=13tw, extraction code: 13tw.
The only difference is, beagle_coco is provided as .zip in Google Drive, while as unzipped folder in Baidu Pan.
Three folders of videos are provided, which all share the same calibration data. All sequences have 10 views. Note that all videos are recorded in linear type using GoPro BLACK 11 cameras.
synced_15m_bk: 15-minute long, 120fps, 1920*1080p, .mp4 format, h264 encoding. Two Beagle dogs. Note that, 6.mp4 was broken during recording, therefore it has only 2min36s long.
synced_15m_1fps: Downsampled as 1fps of the above videos. The first 113 frames are labeled with 3D keypoints and per-view masks. The 0th to 89th frames are used for training 2D detection model, and 90th to 112th frames are used for evaluation. See beagle_coco for cleaned coco dataset.
synced_0.5m_single: 30-second long, 120fps, 1920*1080p, .mp4 format, h264 encoding. Single Beagle dog.
calib: Videos used for calibration (a chessboard with 12*9 grids, grid edge length is 0.03m).
beagle_coco: 2D detection dataset in COCO format. For each image, 2D keypoints and masks are manually labeled. There are 29 keypoints. Note that, image folder images and images_unique have the same images with different names. unique index images from different views in a unified order. In my paper, images_unique is used for training. annotations_unique_train.json is the train part and annotations_unique_eval.json is the evaluation part. In fact, during training 2D pose model, using annotations_unique_eval.json as evaluation may cause the final model test well on 90~112th frames, resulting in a risk of dataset leaking (90th to 112th frames are used for final 3D evaluation). However, as it is only used for generalization demonstration on new species, and consider the limited size of dataset, we ignore such problem. labeled provides the 3D keypoint annotations. For keypoint names, please refer to the json files.
calibdata_beagle: camera_undist.json is the final calibration result. backgrounds have the cleaned background images for all views.
If you find this data useful, please cite the paper
@article{an2023three,
title={Three-dimensional surface motion capture of multiple freely moving pigs using MAMMAL},
author={An, Liang and Ren, Jilong and Yu, Tao and Hai, Tang and Jia, Yichang and Liu, Yebin},
journal={Nature Communications},
volume={14},
number={1},
pages={7727},
year={2023},
publisher={Nature Publishing Group UK London}
}
We thank Beijing Sinogenetic Biotechnology Co., Ltd for the help in the Beagle dog video recording. We also thank Yuxiang Zhang for his help in video recording. Beagle dogs are provided by Beijing Sinogenetic Biotechnology Co., Ltd, and are both treated gentally and carefully.
MIT License
Copyright (c) 2024 An Liang
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.