ABSTRACT
Mass fitness activities are becoming increasingly popular, and it is of great significance to automatically identify fitness exercise categories and counting. Fitness+ artificial intelligence is the future development trend. This paper proposes an integrated method to automatically identify the type of exercise and count the frequency of exercise. On the basis of extracting human joint points, the spatiotemporal graph convolutional network is improved by adding spatial attention modules and temporal dilated convolutional module to identify different types of motion. After identifying the type of motion, the frequency of movement is judged by the changes of the angle characteristics of the human joint point, realizing the integration of fitness activities classification and counting. Finally, the paper conducted experiments on related dataset, where the classification accuracy rate reaches 91.2%, indicating that the network model designed achieved good recognition effects, and the counting accuracy rate reaches 93.4%, indicating the feasibility and effectiveness of the proposed counting method.
- Soro A, Brunner G, Tanner S, Recognition and Repetition Counting for Complex Physical Exercises with Deep Learning[J]. Sensors, 2019, 19(3).Google Scholar
- Li Yanshan, Guo Tianyu, Liu Xing, Skeleton-based Action Recognition with Lie Group and Deep Neural Networks [C]// IEEE International Conference on Signal and Image Processing, China, Wuxi: IEEE press, 2019: 26-30Google Scholar
- Bharath R N, Subramanian A, Ravichandran K, Exploring Techniques to Improve Activity Recognition using Human Pose Skeletons[C]// 2020 IEEE Winter Applications of Computer Vision Workshops (WACVW). IEEE, 2020.Google Scholar
- Yan S, Xiong Y, Lin D . Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition[J]. 2018.Google ScholarCross Ref
- Li, M., : Actional‐structural graph convolutional networks for skeleton‐based action recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 3595–3603. Long Beach (2019)Google Scholar
- Obinata Y, Yamamoto T. Temporal Extension Module for Skeleton-Based Action Recognition [C]// International Conference on Pattern Recognition, Milan, Italy, 2020Google Scholar
- Si Chenyang, Chen Wentao, Wang Wei, An Attention Enhanced Graph Convolutional LSTM Network for Skeleton-Based Action Recognition [C]// IEEE Conference on Computer Vision and Pattern Recognition. USA, CA, Long Beach: IEEE press, 2019: 1227-1236Google Scholar
- Park H J, Baek J W, Kim J H . Imagery based Parametric Classification of Correct and Incorrect Motion for Push-up Counter Using OpenPose [C]// 2020 IEEE 16th International Conference on Automation Science and Engineering (CASE). IEEE, 2020.Google Scholar
- Cao Z, Simon T, Wei S-E, Sheikh Y (2017) Realtime multi-person 2d pose estimation using part affinity fields. In: Computer Vision and Pattern Recognition (CVPR), 2017 9, 10Google Scholar
- Song S, Lan C, Xing J, Zeng W, Liu J (2017) An end-to-end spatio-temporal attention model for human action recognition from skeleton data. In: Proceedings of the thirty-first AAAI conference on artificial intelligence, February 4–9, 2017, San Francisco, California, USA, pp 4263–4270Google ScholarCross Ref
- Kim, T.S., Reiter, A.: Interpretable 3d human action analysis with temporal convolutional networks. In: IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). IEEE, pp. 1623–1631. Honolulu (2017)Google Scholar
Recommendations
The Research of Convolutional Neural Network Based on Integrated Classification in Question Classification
As a new generation of search engine, automatic question answering system (QAS) is becoming more and more important and has become one of the hotspots of computer application research and natural language processing (NLP). However, as an indispensable ...
Improved classification with allocation method and multiple classifiers
We propose a new allocation method for building a classification ensemble.Allocation method uses multiple classifiers: the allocator and micro classifiers.Allocator separates the dataset and allocates them to one of micro classifiers.Allocator is based ...
Fitness diversification in the service of fitness optimization: a comparison study
GECCO '22: Proceedings of the Genetic and Evolutionary Computation Conference CompanionBlindly chasing after fitness is not the best strategy for optimization of hard problems, as it usually leads to premature convergence and getting stuck in low-quality local optima. Several techniques such as niching or quality-diversity algorithms have ...
Comments