HSiPu2 - A New Human Physical Fitness Action Dataset for Recognition and 3D Reconstruction Evaluation

Chuanlei Zhang, Lixin Liu, Minda Yao, Wei Chen, Dufeng Chen, Yuliang Wu; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2021, pp. 3172-3181

Abstract


In this paper a human physical fitness action feature dataset named HSiPu2 is introduced, which contains 8,044 action data sequences and 80,440 images. The dataset is built for three physical fitness actions, which are situp, push-up and pull-up, and each action data has two categories corresponding to standard and non-standard actions. Two cameras work as sensors to capture features from different views. HSiPu2 data set facilitates the evaluation of the performance of machine learning algorithms used in recognition and evaluation problems related to human behaviours recognition. The dataset is freely and publicly available online. A new recognition method based on deep learning techniques, including the twobranch multi-stage convolutional neural networks (CNNs) and long short-term memory (LSTM) networks with attention, are employed to learn the long-term dependencies from videos for human physical fitness action recognition on the HSiPu2. For comparison purpose, traditional machine learning models, including Decision Tree, Random Forest, SVM, Bagging, GBDT, AdaBoost, XGBoost and Voting, are utilized for the action recolonization. Furhtermore, HSiPu2 also can serve as a dataset to evaluating a human action 3D model.

Related Material


[pdf]
[bibtex]
@InProceedings{Zhang_2021_CVPR, author = {Zhang, Chuanlei and Liu, Lixin and Yao, Minda and Chen, Wei and Chen, Dufeng and Wu, Yuliang}, title = {HSiPu2 - A New Human Physical Fitness Action Dataset for Recognition and 3D Reconstruction Evaluation}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops}, month = {June}, year = {2021}, pages = {3172-3181} }