Identifying weapon-carrying actions in video surveillance systems through self-attention

Jesús Ruiz-Santaquiteria

doi:10.1016/j.imavis.2026.106087

Identifying weapon-carrying actions in video surveillance systems through self-attention

2026-06-19

0

OA

AI

J

Jesús Ruiz-Santaquiteria *

O

Oscar Deniz

G

Gloria Bueno

DOI：10.1016/j.imavis.2026.106087

Original

Original request for help

Save

Abstract

En 中文

• A new architecture for action recognition in video surveillance systems is proposed. • The method is based on a Transformer encoder module and a combination of 2D keypoints. • A new action recognition dataset for gun-carrying actions is also proposed. • Experiments show that the proposed method outperforms other similar methods.

Keywords:

Human action recognition

CCTV surveillance

Weapon detection

Human pose estimation

Identifying weapon-carrying actions in video surveillance systems through self-attention

Abstract

Journal

Organization

Cited Papers

Citing Papers