Download

Please send an e-mail to yanmnn@stu.xmu.edu.cn, including contact details (title, full name, organization, and country) and the purpose for downloading the dataset. By sending the e-mail you accept the following License.

Note: Due to the open source license restrictions of the SMPL, we cannot directly provide .pkl and .npz files. If you have questions about missing files, please contact the first author.

Dataset Explanation

  1. For the convenience of training, we put all the data in Dataset_Train and Dataset_Test in this release of AscendMotion. In order to be compatible with other LiDARHumanMotion works (for example, http://www.lidarhumanmotion.net/sloper4d/), we changed the data format to be consistent with the series work so that you can use the same Dataloader. In \AscendMotion_Dataset_Release_v1\Dataset_Train, you will find two files, .npz and .pkl. The .npz file is mainly used for training and evaluation of monocular camera-based methods, and the .pkl file is mainly used for training and evaluation of radar and multi-modal methods.

  2. The bvh file in each subfile is the source file of the IMU acquisition system. If you read this file directly, you can see that bvh will drift over time, which is an obvious defect of the IMU acquisition device. But don’t worry, the data in the .npz and .pkl files mentioned above have solved this problem. You can read the data in these two files directly. The bvh file only provides reference values ​​for the source file. In particular, if you need more additional IMU detailed information (such as timestamps, joint velocity acceleration, etc.), you can find it from the bvh file.

  3. The results of the high-precision human SMPL scans of each coach are in the body_smpl.json file of each scene sequence.

  4. Finally, regarding visualization, we open source the registered rock wall scenes, including mesh format and ultra-high precision point cloud format. The files are in each separate scene sequence, such as \AscendMotion_Dataset_Release_v1\20241016JimeiYanwu\WL. You can directly put the human body in the pkl sequence into the coordinate system consistent with the scene to achieve the visualization effect in the paper.

Citation

@InProceedings{yan2025climbingcap,
    author    = {Yan, Ming and Lin, Xincheng and Luo, Yuhua and Fan, Shuqi and Dai, Yudi and Zhong, Qixin and Zhong, Lincai and Ma, Yuexin and Xu, Lan and Wen, Chenglu and Shen, Siqi and Wang, Cheng},
    title     = {ClimbingCap: Multi-Modal Dataset and Method for Rock Climbing in World Coordinate},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2025},
}

Commercial licensing

Please email us:
siqishen@xmu.edu.cn
clwen@xmu.edu.cn
cwang@xmu.edu.cn