Temporal Segmentation of Fine-grained Semantic action: A Motion-Centered Figure Skating Dataset
Shenglan Liu
Aibin Zhang
Yunheng Li
Jian Zhou
Li Xu
Zhuben Dong
Renhao Zhang
Dalian University of Technology




A video in MCFS. The labels of this video belong to the subset-level.


Abstract

Temporal Action Segmentation (TAS) has achieved great success in many fields such as exercise rehabilitation, movie editing, etc. Currently, task-driven TAS is a central topic in human action analysis. However, motion-centered TAS, as an important topic, is little researched due to unavailable datasets. In order to explore more models and practical applications of motion-centered TAS, we introduce a Motion-Centered Figure Skating (MCFS) dataset in this paper. Compared with existing temporal action segmentation datasets, the MCFS dataset is fine-grained semantic, specialized and motion-centered. Besides, RGB-based and Skeleton-based features are provided in the MCFS dataset. Experimental results show that existing state-of-the-art methods are difficult to achieve excellent segmentation results (including accuracy, edit and F1 score) in the MCFS dataset. This indicates that MCFS is a challenging dataset for motion-centered TAS. The latest dataset can be ed at https://shenglanliu.github.io/mcfs-dataset/.


Demo video

Temporal Action Segmentation Video（MCFS)


Dataset hierarchy
A three level semantics annotations and collect 4 sets (e.g. Spin), 22 subsets (e.g. CamelSpin) and 130 elements (e.g. CamelSpin3) at each annotation level.


Experiments

(1) Comparison with the state-of-the-art on 50Salads, GTEA, and the Breakfast dataset. (* All data obtained from (Farha and Gall 2019) and (Chen et al. 2020a)).


Comparison with the state-of-the-art on 50Salads, GTEA, and the Breakfast dataset. (* All data obtained from (Farha and Gall 2019) and (Chen et al. 2020a)).

(2) Element-level action recognition results of representative methods. Specifically, results of recognizing element categories across all set, within a subset, and within an element.


Element-level action recognition results of representative methods. Specifically, results of recognizing element categories across all set, within a subset, and within an element.

(3) Qualitative results for the TAS task on MCFS.


Qualitative results for the TAS task on MCFS.



Download

MCFS



All MCFS Dataset(code:yhfc)

Paper

Liu, Zhang, Li
Temporal Segmentation of Fine-grained Semantic action: A Motion-Centered Figure Skating Dataset
(Paper)
(Additional details/
supplementary materials)


Acknowledgements

We sincerely thank the outstanding annotation team for their excellent work. The template of this webpage is borrowed from Dian Shao.


Contact

For further questions and suggestions, please contact Shenglan Liu (liusl@mail.dlut.edu.cn).