Research Area: Machine Learning
In order to achieve accurate judgment and identification of group behaviors, the hierarchical deep network model is constructed to judge the group behaviors. Through the construction of the hierarchical deep network model, the group behaviors are judged; the stability, accuracy, expression movement, orientation, error, and work efficiency of the model, as well as the support vector machine model and the convolution network model, are compared and analyzed. The hierarchical depth network model has distinct advantages in comparison with the support vector machine model and the convolution network model. The standard deviation of the hierarchical depth network model is 0.013, while the standard deviations of the other two models are larger than the hierarchical depth network model. The proposed algorithm has an excellent performance in detection accuracy and error. Compared with the other two models, it has certain advantages. In addition, the hierarchical depth network is used for recognizing human behaviors and orientations, as well as extracting and recovering the expressions in group behaviors. Compared with the other two models, the proposed model is also more efficient. The model used in this study has little influence on group behavior recognition by the regional environment and other factors, and there is no significant difference in the judgment results. The operation of the model is studied by identifying the group behaviors based on the hierarchical depth network. The research results show that the model proposed in this study has a comprehensive and excellent result, which also indicates that group behavior recognition is an overall result. It is necessary to have an accurate identification of multiple layer parameters. The research in this study has greatly improved the understanding of hierarchical deep network and group behavior recognition.
Keywords:
Deep hierarchical network
Group
Behavior recognition
Efficiency
Extraction
Deep Learning
Machine Learning
Author(s) Name: Shuhan Qiao, Lukun Wang & Zhiyong Gao
Journal name: Neural Computing and Applications
Conferrence name:
Publisher name: Springer
DOI: 10.1007/s00521-019-04699-4
Volume Information: volume 32, pages5389–5398 (2020)
