Research on Weapon Recognition of Human Body Based on Deep Learning
HUANG Zhaonian
LU Longsheng
CHENG Peng
LI Heng
Abstract：This paper proposes a method based on the improved ST-GCN and YOLOv5 to identify whether a person is carrying a weapon.In this method,a single frame is extracted from the border surveillance video as a unit to extract the skeleton point infor-mation of the human body,and then multi-frame image information is aggregated by using the ST-GCN as a framework to identify the movement of the person.Then,the relationship between personnel and weapons can be detected by YOLOv5 and human skele-ton to determine whether personnel in motion carry weapons.Finally,experiments are conducted to verify the effectiveness of the method.The results show that the method can make full use of the spatio-temporal information between the skeleton points in multi-frame images to accurately identify the movement of people and whether they carry weapons,with good accuracy and robust-ness.
Keywords：action recognitionweapon recognitionskeleton information extractionST-GCN
Publication Date：2025-03-20
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 38-42 )