Warning of Target Attitude of Crowd in Closed Area Based on CNN and Clustering Algorithm

Lianwen Lu, Yuhan Cui, Xinghua Lu

2022

Abstract

Human Activity Recognition (HAR) technology is a hotspot in the field of computer vision. There are still many technical difficulties in the collaborative tracking of human and object. Based on the skeletal point algorithm and target detection and tracking, this paper attempts to build a new cooperative tracking system between people and objects in a relatively closed environment to manage small and medium-sized populations. Label training is carried out for specific posture and specific dangerous goods, so as to realize early warning ability by identifying multi person posture and dangerous goods. Using multi label classification to mark a category can improve recognition efficiency and flexibility, and avoid absolute interpretation in target detection. After multi label training, the specific target object and target pose can improve the accuracy of interactive recognition between human pose and object in real scene. In this paper, we use convolution neural network and clustering algorithm, C3d two stream, openpose human feature bone point recognition model and yolov4 to realize the crowd target attitude early warning in closed area. The final clustering test shows that the proposed method can improve the efficiency of machine learning, enhance the robustness, and improve the accuracy of target attitude.

Download


Paper Citation


in Harvard Style

Lu L., Cui Y. and Lu X. (2022). Warning of Target Attitude of Crowd in Closed Area Based on CNN and Clustering Algorithm. In Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology - Volume 1: ICPDI; ISBN 978-989-758-620-0, SciTePress, pages 647-658. DOI: 10.5220/0011753700003607


in Bibtex Style

@conference{icpdi22,
author={Lianwen Lu and Yuhan Cui and Xinghua Lu},
title={Warning of Target Attitude of Crowd in Closed Area Based on CNN and Clustering Algorithm},
booktitle={Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology - Volume 1: ICPDI},
year={2022},
pages={647-658},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0011753700003607},
isbn={978-989-758-620-0},
}


in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Public Management, Digital Economy and Internet Technology - Volume 1: ICPDI
TI - Warning of Target Attitude of Crowd in Closed Area Based on CNN and Clustering Algorithm
SN - 978-989-758-620-0
AU - Lu L.
AU - Cui Y.
AU - Lu X.
PY - 2022
SP - 647
EP - 658
DO - 10.5220/0011753700003607
PB - SciTePress