Handgun Detection Using Combined Human Pose and Weapon Appearance

Abstract

Closed-circuit television (CCTV) systems are essential nowadays to prevent security threats or dangerous situations, in which early detection is crucial. Novel deep learning-based methods have allowed to develop automatic weapon detectors with promising results. However, these approaches are mainly based on visual weapon appearance only. For handguns, body pose may be a useful cue, especially in cases where the gun is barely visible. In this work, a novel method is proposed to combine, in a single architecture, both weapon appearance and human pose information. First, pose keypoints are estimated to extract hand regions and generate binary pose images, which are the model inputs. Then, each input is processed in different subnetworks and combined to produce the handgun bounding box. Results obtained show that the combined model improves the handgun detection state of the art, achieving from 4.23 to 18.9 AP points more than the best previous approach. INDEX TERMSCCTV surveillance, deep learning, handgun detection, human pose estimation.

Street-crimes modelled arms recognition technique employing deep learning and quantum deep learning (SMARTED)

“…In this paper, the study started from the manual approach which includes the weighted mixture of Gaussians [2], polarized signals-based technique, a multiresolution mosaic technique, and three-dimensional (3D) computed tomography (CT) [3] Haar cascades procedure. Advancements began with machine learning Nowadays, Santaquiteria et al [6] preventing security risks is critical, and deep learning-based technologies enable the development of automatic weapon detectors. Muchiri et al [7] presented the approach to combining information on weapon appearance and human position.…”

“…The backbone can be built using traditional CNN designs like ResNet50 or ResNet101. In the backbone, RPN (region proposal network) [42], feature pyramids [32], and FPN [6] are used (as discussed above).…”

Weapon Detection in Surveillance Videos Using Human Inspired Particle Swarm Optimization Algorithm and Support Vector Machine

“…As stated previously, the YOLO V3 detector needs high intensive graphics processing units and more computation resources for data training. J. Ruiz-Santaquiteria [20] combined both weapon appearance and human pose information for handgun detection. However, the developed model showed only comparable results in the factors like camera distance, poor occlusions, and lighting conditions.…”

“…( 19), so that the problem of optimization is solved using Lagrange function 𝜗 𝑖 , which is mathematically expressed in Eq. (20). (20) that decreases the computational complexity in high dimensional data.…”

“…Experimental results of HIPSO-SVM model on the gun movies database 40% of precision on YouTube database, and 44.10% of recall and 98.40% of precision on Gun movies database. Further, Ruiz-Santaquiteria[20] integrated weapon appearance and human pose information for effective handgun detection. Experimental results confirmed that the presented model obtained 83.83% of recall and 97.3% of precision on YouTube database.…”