Fig 1 - uploaded by Vasileios Mavroeidis
Content may be subject to copyright.
Source publication
In recent years we have seen an upsurge in terror attacks around the world. Such attacks usually happen in public places with large crowds to cause the most damage possible and get the most attention. Even though surveillance cameras are assumed to be a powerful tool, their effect in preventing crime is far from clear due to either limitation in th...
Context in source publication
Context 1
... is worth noting that this type of firearm in addition to other rifles was used in terror attacks (mass shooting), such as in New Zealand in 2018 [20], in Las Vegas, USA in 2017 [21] and in Orlando, USA in 2016 [22]. We identified four essential component parts: barrel, magazine, butt-stock and upper-lower receiver assembly (see Figure 1). To generate positive instances for our dataset we queried the Google images service and retrieved 4500 instances of pictures depicting AR-15 rifles. ...
Similar publications
In recent years, the number of Gun-related incidents has crossed over 250,000 per year and over 85% of the existing 1 billion firearms are in civilian hands, manual monitoring has not proven effective in detecting firearms. which is why an automated weapon detection system is needed. Various automated convolutional neural networks (CNN) weapon dete...
Citations
... At most, each video frame may first be preprocessed to highlight any human bodies visible in the image, using a separate, pretrained DNN for person detection [47] or for human body pose estimation [48] [49]. A notable example that deviates from the norm is the method in [50], which exploits an ensemble of multiple, simple CNNs, instead of one monolithic DNN, for weapon detection in images. Each model detects a specific firearm component/part (e.g., barrel) and their outputs are aggregated to obtain a single final prediction. ...
p>Recent trends in the modus operandi of technologically-aware criminal groups engaged in illicit goods trafficking (e.g., firearms, drugs, cultural artifacts, etc.) have given rise to significant security challenges. The use of cryptocurrency-based payments, 3D printing, social media and/or the Dark Web by organized crime leads to transactions beyond the reach of authorities, thus opening up new business opportunities to criminal actors at the expense of the greater societal good and the rule of law. As a result, a lot of scientific effort has been expended on handling these challenges, with Artificial Intelligence (AI) at the forefront of this quest; mostly machine learning and data mining methods that can automate large-scale information analysis. Deep Neural Networks (DNNs) and graph analytics have been employed for automatically monitoring and analyzing the digital activities of large criminal networks in a data-driven manner. However, such practices unavoidably give rise to ethical and legal issues, which need to be properly considered and mitigated. This paper is the first one to systematically explore all of the above aspects jointly and from a combined perspective, without focusing on a particular angle or type of illicit goods trafficking. It emphasizes how advances in AI both allow the authorities to unravel technologically-aware trafficking networks and provide countermeasures against any potential violations of citizen rights in the name of security.</p
... At most, each video frame may first be preprocessed to highlight any human bodies visible in the image, using a separate, pretrained DNN for person detection [47] or for human body pose estimation [48] [49]. A notable example that deviates from the norm is the method in [50], which exploits an ensemble of multiple, simple CNNs, instead of one monolithic DNN, for weapon detection in images. Each model detects a specific firearm component/part (e.g., barrel) and their outputs are aggregated to obtain a single final prediction. ...
p>Recent trends in the modus operandi of technologically-aware criminal groups engaged in illicit goods trafficking (e.g., firearms, drugs, cultural artifacts, etc.) have given rise to significant security challenges. The use of cryptocurrency-based payments, 3D printing, social media and/or the Dark Web by organized crime leads to transactions beyond the reach of authorities, thus opening up new business opportunities to criminal actors at the expense of the greater societal good and the rule of law. As a result, a lot of scientific effort has been expended on handling these challenges, with Artificial Intelligence (AI) at the forefront of this quest; mostly machine learning and data mining methods that can automate large-scale information analysis. Deep Neural Networks (DNNs) and graph analytics have been employed for automatically monitoring and analyzing the digital activities of large criminal networks in a data-driven manner. However, such practices unavoidably give rise to ethical and legal issues, which need to be properly considered and mitigated. This paper is the first one to systematically explore all of the above aspects jointly and from a combined perspective, without focusing on a particular angle or type of illicit goods trafficking. It emphasizes how advances in AI both allow the authorities to unravel technologically-aware trafficking networks and provide countermeasures against any potential violations of citizen rights in the name of security.</p
... Then, the proposed deep neural network is used to detect and locate the gun. Eventually, the outcome has been obtained through the aggregation of its different outcomes [16]. Here the issue is that the authors use only one type of rifle, i.e., the AR-15. ...
... The classification results are evaluated based on accuracy, area under the curve and F1-Score metrics. The performances of all the feature selection techniques for classifiers have been given in Tables 13,14,15,16,17 and 18. The visual representations of the proposed outputs are shown in Fig. 10. Figure 10(a)-(c) indicate hand poses. ...
Violence due to firearms is a menace and is growing across the globe. Mostly small firearms such as pistols and revolvers are used in close proximity for violence in public places. Our moral duty is to contribute to solving this problem with our expertise as it is impossible to keep 24 ×\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\times $$\end{document} 7 eyes on everyone in the street and corners. An automatic lightweight artificial intelligence solution to detect such small firearms is a welcoming step towards a more secure society. Here, we have proposed a hand pose pattern analysis to identify whether a person is holding a pistol or revolver. Furthermore, different data prepossessing, machine learning and deep learning techniques have been implemented to classify guns and no-gun from visual media. Finally, we have introduced a metric learning approach instead of classification along with our novel Fuzzy Discernible Feature Selection (FDFS) technique to provide faster and more accurate discrimination between gun and non-gun instances from images and videos. The result from Deep Neural Network + FDFS has achieved 93% test accuracy and outperforms the other used machine learning and deep learning alternatives.
... However, the task of monitoring video surveillance is both time-consuming and tedious. Egiazarov et al. (2019) have underscored the limitations of security cameras in deterring and identifying crimes, citing a rise in global terrorist attacks. To tackle this issue, they proposed a firearm detection system based on a semantic CNN grouping for the various components of a weapon. ...
The popularity of social networks and video-on-demand platforms has increased the importance of image and video censorship. These platforms contain various content such as violence, explicit content, drug use and smoking that may be offensive or harmful to certain viewers. As a result, censorship is employed to filter or remove content that is unsuitable for particular audiences such as children and teenagers. However, policies for censoring harmful content in digital environments are either limited or nonexistent. This underscores the need for automated systems that can detect and censor harmful content in real time. To address these challenges, we developed the first of our knowledge systems using deep learning techniques to censor harmful content. We propose two novel, YOLO-based real-time censorship algorithms. Our approaches employ a pipeline-based architecture that parallelizes the operations with subprocesses. In our experiments, the proposed algorithms performed faster and with higher accuracies compared to traditional approaches. Specifically, its content-based accuracies were 98% for explicit content, 97% for alcohol, 98% for cigarettes and 97% for violence. Our research highlights the importance of developing effective and efficient solutions for censoring harmful content on digital media platforms. Our deep learning-based system represents a promising approach to this challenge and has the potential to enhance user safety and protect vulnerable groups from harmful and offensive content. Future research will continue to refine and improve such systems to better address the evolving landscape of digital media and the challenges posed by harmful content.
... The comes about are analyzed after preparing and testing models on distinctive datasets [18]. Several researchers have done firearm detection using fuzzy classifiers [8] and an ensemble of neural networks [9]. ...
In this day and age, we are witness to ever increasing gun violence all around the world. Technology has surpassed all human beliefs where each person can be easily tracked through their mobiles or through the fortitude of CCTV cameras available all across public properties and areas. There is a need to stop gun violence to protect people's Right to Live. There are several instances appearing in the news daily about deaths caused due to gun violence. An alarm based system can be introduced which tracks the publicly available CCTV footage to look for guns in the open and raise appropriate alarms. In order to achieve this a robust model to identify and classify firearms automatically from videos is required. The aim of this paper is to describe a YOLO-based model which is highly effective in recognizing firearms in videos and mark them in the video such that the model can be further used for further applications such as raising alarms, tracking human beings with firearms etc.
... However, profound learning-based gun detection studies are relatively fewer regarding the number of research papers, specifically with face detection-some of the scholarly articles where focus on guns discussed in this section. output has been achieved using an aggregation of its different outcomes [27]. The research is restricted to only a single type of rifle, which is the AR-15. ...
Violence, in any form, is a disgrace to our civilized world. Nevertheless, even in modern days, violence is an integral part of our society and causes the deaths of many innocent lives. One of the conventional means of violence is using a firearm. Firearm-related death is currently a global phenomenon. It is a threat to society and a challenge to law enforcement agencies. A significant portion of such crimes happens in semi-urban areas or cities. Nowadays, CCTV-based surveillance is widely used by governments and private organizations for monitoring and prevention. However, human-based monitoring requires a significant amount of person-hours as a resource and is prone to mistakes. On the other hand, automated smart surveillance for violent activities is more suitable for scale and reliability. The paper’s main focus is to showcase that deep learning-based techniques can be used in combination to detect firearms (particularly guns). This paper uses different detection techniques, such as Faster Region-Based Convolutional Neural Networks (Faster RCNN) and the latest EfficientDet-based architectures for detecting guns and human faces. An ensemble (stacked) scheme has improved the detection performance to identify human faces and guns at the post-processing level using Non-Maximum Suppression, Non-Maximum Weighted, and Weighted Boxes Fusion techniques. This paper has empirically discussed the comparative results of various detection techniques and their ensembles. It helps the police to gather quick intelligence about the incident and take preventive measures at the earliest. Also, the same method can be used to identify social-media videos for gun-based content detection. Here, theWeighted Boxes Fusion-based ensemble detection scheme provides mean average precision 77.02%, 16.40%, 29.73% for the mAP
<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.5</sub>
, mAP
<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">0.75</sub>
and mAP
<sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">[0.500.95]</sub>
, respectively. The results achieve the best performance among all the experimented alternatives. The model has been rigorously tested with unknown test images and movie clips. The obtained ensemble schemes are satisfactory and consistently improve over primary models.
... 6 Firearm Detection and Segmentation Using an Ensemble of Semantic Neural Networks Come up with a system of recognition of weapon along with its positioning through a Semantic Neural Network. Development needs to be done to train the system with further rifle's constituent dataset [10]. 7 Identifications of a concealed weapon in a Human Body Newly proposed fusion algorithm of RGB with its corresponding Infrared image by capturing emitting radiation. ...
This paper discusses the approach that discovers the suspiciousness of a suspected person and also perceives whether the suspicious person has concealed (hidden) firearm or not as well as evaluates its accuracy with the testimony of suspiciousness from IR image accurately and efficiently in order to save time in the confirmation of suspect while breaking down the barrier with the assurance by spotting the hidden weapon under suspected suspicious person's clothes with supervised machine learning, 2D convolutional neural network approach with a 3-by-3 layer for the intention of the detection of weapon and suspicious assessment measured under Digital Forensics.
... In this section, we will discuss the related approaches relevant to our work. In 2020, Egiazarov et al. [12] present a model for weapon detection and segmentation by using semantic segmentation model. They have performed their experiments on AR-15 Rifle dataset. ...
In the last few years, terror activities across the world have been raised drastically. Therefore, we need a framework that can detect these terror and illegal actions automatically. In spite of various state-of-the-art deep learning algorithms, weapon detection is still a serious challenge. This work focuses on the classification and detection of guns which integrates deep fusion of feature information to generate the most discriminant feature vector. Firstly, we utilize the fully connected layers of recent deep CNN models: Inception-ResNetv2 and MobileNetv2 for feature extraction, and then we fuse these features by using concatenation operation. To acquire a more compact presentation of features and reduce the complexity of computation, we have utilized NCA. After that, we classify the images by using an SVM classifier. Finally, to detect the guns in an image, a bounding box regression module is proposed by applying LSTM. Qualitative and quantitative outcomes indicate that our framework can detect guns with huge variations in size along with rigorous occlusion. In particular, our method achieves an accuracy rate of 98.62% with a false alarm rate of 1.38% on the gun dataset which is extracted from Kaggle.
... accuracy for their proposed work. Egiazarov et al. [29] presented a weapon detection system using an ensemble of semantic CNNs. They claimed that their proposed approach decomposes the problem of detecting and locating a weapon into a set of smaller problems, concerned with the individual component parts of a weapon. ...
The use of weapons nowadays is becoming a leading cause of severe crimes in our society, which reluctantly results in dreadful consequences. The weapons used typically varies from knife, iron-rod, dagger, sabre to firearms like guns and bombs. Due to the unavailability of any proactive mechanism for avoiding heinous crimes using such weapons, an active surveillance performing real time weapon identification is proposed here, as a boon to societal security requirement. As part of it, this paper presents a novel approach based on Convolutional Neural Network (CNN) for identifying visual weapons. This proposed CNN model is initialized with the pre-trained Visual Geometry Group-16 (VGG-16) network weights. These weights are further fine-tuned by training this CNN model with comprehensive weapons (knives and handguns only) and non-weapon images. Weapon category images correspond to further classified classes of “isolated” and “handheld” weapons. However, weapon identification is challenging because of unavailability of diverse databases containing images with variations in shape, texture, scale, occlusion of weapon, etc. This paper reduces this limitation by presenting an algorithm for generating new images and other algorithm for preprocessing the images for quality enhancement. The accuracy achieved is 98.07% with original isolated images and 98.36% with its preprocessed images, while 98.42% with original handheld images and 98.80% with its preprocessed images. The preprocessed algorithm’s applicability is confirmed by the higher accuracy achieved by this model using preprocessed images. The accuracy achieved is on an average of ~7% higher than those achieved by other researchers with similar work. The improved result of weapon identification in terms of accuracy proves the appropriateness of the proposed research in being used commercially.
... Now, a simple deep neural network can easily detect and locate the weapon. Finally, the final output has been achieved using an aggregation of its different outcomes [15]. The research is restricted to only a single type of rifle, which is AR-15. ...
... 6. [11] BrandImageNet to detect brands in social media images. BrandImageNet is a modified Berkeley Vision and Learning Center (BVLC) reference CaffeNet model [12] Deep Belief Network (DBN) and Long Short-Term Memory (LSTM) for predicting potential road accidents [13] Remove the redundant or irrelevant (noisy) images captured by a fine tuned VGG16 network from a disaster-hit area to take effective decision-making [14] Deep learning technique to mine geo-tagged Instagram images for a reliable indirect source to examine lifestyle related diseases [15] An ensemble deep learning method to decompose a gun image into different parts for better detection [16] Transfer learning technique has been used for gun detection from X-Ray baggage images [17] Faster RCNN with VGG16 backbone network is used to detect a gun (pistol) from five consecutive frames of a given input video and set a alarm if one detected 2 Conceptual overview ...
A large portion of the global population generates various multimedia data such as texts, images, videos, etc. One of the
most common categories which infuences the public at large is visual multimedia content. Due to the diferent social media
platforms (e.g., Whatsapp, Twitter, Facebook, Instagram, and YouTube), these materials are passed without censorship and
national boundaries. Multimedia data containing any violent or vulgar objects could trigger public unrest, and thus, it is a
serious threat to the law and order of the land. Children and teenagers use social media like never before in previous generations and create lots of multimedia data. It is important to assess the quality of multimedia content without any bias and
prejudices. Although the mainstream social media platforms use diferent flters and moderation using human experts, it is
impossible to verify the terabytes of uploaded images and videos. Thus, it is inevitable to automate the content assessment
phase without incurring an increase in upload time. This study aims to prevent uploading or to tag an image/video with a
reasonable percentage of a gun as content. In this paper, object detection architectures such as Faster RCNN, EfcientDet,
and YOLOv5 have been used to demonstrate how these techniques can efciently detect human faces and diferent types of
guns in given multimedia data (images/videos). The models are tested on various test images and video clips. A comparative
analysis has also been discussed based on mean average precision and frames per second metric. The YOLOv5 provides
the best-performing results as high as 80.39% and 35.22% at mAP0.5 and mAP[0.50∶0.95]
, respectively. A face recognition task
requires thousands of samples and the usual deep learning models are data-driven. On the contrary, a few-shot learning
approach has been implemented to recognize the detected faces categorizing the content as real or reel.
