Group Activity Recognition Based on GFU and Hierarchical LSTM

WANG Chuan-xu; XUE Hao

doi:10.3969/j.issn.0372-2112.2020.08.002

您当前的位置：

首页 >

文章列表页 >

Group Activity Recognition Based on GFU and Hierarchical LSTM

更新时间：2025-07-16

- Group Activity Recognition Based on GFU and Hierarchical LSTM
- Acta Electronica Sinica Vol. 48, Issue 8, Pages: 1465-1471(2020)
- 作者机构：
  
  青岛科技大学信息科学技术学院,山东,青岛,266001
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China (No.61672305)
- DOI：10.3969/j.issn.0372-2112.2020.08.002
  CLC： TP301.6
- Published Online：25 August 2020，
  
  Published：2020
- 稿件说明：
移动端阅览
WANG Chuan-xu, XUE Hao. Group Activity Recognition Based on GFU and Hierarchical LSTM[J]. Acta Electronica Sinica, 2020, 48(8): 1465-1471.
DOI：

WANG Chuan-xu, XUE Hao. Group Activity Recognition Based on GFU and Hierarchical LSTM[J]. Acta Electronica Sinica, 2020, 48(8): 1465-1471. DOI： 10.3969/j.issn.0372-2112.2020.08.002.

摘要

提出一种以"关键人物"为核心，使用门控融合单元（GFU，Gated Fusion Unit）进行特征融合的组群行为识别框架，旨在解决两个问题：①组群行为信息冗余，重点关注关键人物行为特征，忽略无关人员对组群行为的影响；②组群内部交互行为复杂，使用GFU有效融合以关键人物为核心的交互特征，再通过LSTM时序建模成为表征能力更强的组群特征.最终，通过softmax分类器进行组群行为类别分类.该算法在排球数据集上取得了86.7%的平均识别率.

Abstract

This paper proposes a group behavior recognition framework with "key persons" as the core and Gated Fusing Unit (GFU) for feature fusion. Its aim is to solve the following two problems: 1) Group behavior information is redundant

focusing on key person behavior characteristics

ignoring the influence of unrelated personels on group behavior. 2)The internal interaction relationship is complex within group

GFU is used to effectively model interaction feature centered around the key characters and it is temporally evolved into the group characteristics via LSTM processing. Finally

the group behavior category is classified with Softmax. The algorithm achieves an average recognition rate of 86.7% on the volleyball dataset.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Summarization of Group Activity Recognition Algorithms Based on Deep Learning Frame

Research on Group Behavior Recognition Method Based on Multi-Stream Architecture and Long Short-Term Memory Network

Related Author

DENG Hai-gang

WANG Chuan-xu

LI Cheng-wei

LIN Xiao-meng

HU Xiao-yue

MENG Wei-jia

YAN Chun-juan

Hai-gang DENG

Related Institution

School of Instrumentation Science and Engineering, Harbin Institute of Technology

School of Information Science and Technology, Qingdao University of Science and Technology

School of Instrumentation Science and Engineering， Harbin Institute of Technology

School of Information Science and Technology， Qingdao University of Science and Technology

⁰