Page 2440
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Weapon Detection and Alert System Using Yolo Deep Learning
Technique
Dr Santhosh Kumar B N
1
, Dr H S Nagalakshmi
2
, Dr Prakasha Raje Urs
3
1
Associate Professor, Department of Computer Science, Maharani’s Science College for Women
(Autonomous), Mysore, Karnataka, India.
2
Associate Professor and Head, Department of BCA, Government College for Women (Autonomous),
Mandya, Karnataka, India.
3
Associate Professor, Department of Computer Science, Government First Grade College , Nanjanagud,
Karnataka, India.
DOI:
https://doi.org/10.51583/IJLTEMAS.2026.150500195
Received: 10 May 2026; Accepted: 15 May 2026; Published: 12 June 2026
ABSTRACT
Gun-related violence poses a major threat to public safety and well-being worldwide. Traditional surveillance
systems rely on human monitoring, which is prone to fatigue and delayed response. This paper proposes a smart
surveillance system for real-time weapon detection using deep learning techniques. The system employs the
YOLOv3 algorithm to detect guns, rifles, and fire from video streams with high accuracy. It also captures the
location of incidents and stores data for further analysis. A multi-system architecture is implemented using socket
programming to simulate real-world integration. The proposed model ensures fast detection with reduced
computational cost. It enhances response time and minimizes human dependency in surveillance tasks. The
system can be extended to autonomous security and robotic applications. Experimental results demonstrate its
effectiveness in improving public safety and security systems.
Keywords: YOLO, Deep Learning, Weapon Detection, Computer Vision, Surveillance System
INTRODUCTION
Gun-related violence is a major concern affecting public safety, health, and economic stability across the world
A significant number of deaths occur each year due to firearm-related incidents.
Exposure to such violence often leads to psychological trauma, especially among children.
Children who witness or experience violence may suffer from long-term mental health issues.
Handheld guns are commonly used in crimes such as robbery, burglary, and assault.
Early detection of suspicious activities can help reduce such crimes effectively.
Traditional surveillance systems depend on human monitoring, which is prone to fatigue and delays.
Continuous observation over long periods reduces attention and increases chances of missing critical events.
Advancements in machine learning and computer vision have enabled automated surveillance systems. Deep
learning techniques, especially convolutional neural networks, improve object detection accuracy.
The availability of large datasets and powerful GPUs has accelerated these developments.
This research proposes a smart surveillance system for real-time weapon detection.
The system uses the YOLOv3 algorithm to detect guns, rifles, and fire efficiently.
It also identifies incident locations and stores data for further analysis.
Page 2441
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
The proposed system enhances security, reduces human effort, and supports future robotic surveillance
applications. The developed system aims to minimize computational complexity while maintaining real-time
performance, making it suitable for deployment in smart surveillance environments. Furthermore, the proposed
approach can be extended to autonomous security systems and surveillance robots for enhanced threat detection.
Experimental results demonstrate the effectiveness of the system in improving response time and reducing
reliance on manual monitoring, thereby contributing to safer and more efficient security infrastructures.
Objectives
The main objectives of the proposed system are:
Real-time detection of weapons from live camera feeds
High detection accuracy with minimal false positives
Automatic image capture upon detection
Alert generation via buzzer and email notification
Storage of detection data for future analysis
LITERATURE SURVEY
Several studies have explored weapon detection using deep learning:
YOLOv3-based systems provide high accuracy and faster detection compared to earlier models.
YOLOv4 combined with CNN improves detection efficiency and supports email alerts.
YOLOv8-based systems achieve improved Mean Average Precision (mAP ≈ 0.78) and real-time usability.
These works highlight the effectiveness of YOLO models in surveillance applications.
System Architecture Diagram
Page 2442
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
Description
The system architecture consists of a camera capturing live video, which is processed by the YOLO model.
Detected weapons trigger alert mechanisms including buzzer activation and email notifications. The system also
stores captured images for further analysis.
Flowchart of Proposed System
Page 2444
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
2. Capture video frames
3. Load YOLO model
4. Detect weapon
5. If detected → Trigger alert & send email
6. Else → Continue monitoring
Data Flow Diagram (DFD)
Description:
The DFD shows how input frames from the camera are converted into feature maps, processed by the YOLO
model, and output as detected weapon classes with alerts.
Page 2445
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
METHODOLOGY
YOLO Algorithm
YOLO (You Only Look Once) is a real-time object detection algorithm that treats detection as a regression
problem. It divides images into grids and predicts bounding boxes and class probabilities.
Key Techniques
Bounding Box Detection: Identifies object location
Intersection over Union (IoU): Measures detection accuracy
Residual Blocks: Improve deep learning performance
CNN Integration
Convolutional Neural Networks (CNNs) extract image features and improve detection accuracy by identifying
patterns such as edges and shapes.
System Design
Architecture
Camera captures live feed
Frames processed using YOLO model
Weapon detection performed
Alert system triggered
Image sent via email
Workflow
1. Start system
2. Capture video frames
3. Load trained YOLO model
4. Detect weapon
5. If detected:
Trigger buzzer
Capture image
Send email alert
Stop system
Page 2446
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
YOLO Detection Process
Description:YOLO divides an image into grids and predicts bounding boxes with confidence scores.
Intersection over Union (IoU) ensures accurate localization of detected objects.
Page 2447
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
RESULTS AND DISCUSSION
Performance Graphs
Description:
Precision-Recall Curve: Shows trade-off between detection accuracy and recall
mAP Graph: Evaluates overall model performance
Loss vs Epochs: Demonstrates training convergence
Confusion Matrix: Displays classification accuracy
Page 2448
www.rsisinternational.org
INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)
ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XV, Issue V, May 2026
RESULTS AND DISCUSSION
The system successfully detects weapons in real-time with high accuracy. The integration of alert mechanisms
ensures immediate response, reducing potential threats.
Key observations:
Fast detection speed
Reliable alert system
Reduced human dependency
Advantages
Real-time monitoring
Automated alerts
High accuracy
Reduced human error
Scalable for large systems
Limitations
Performance depends on lighting conditions
Requires computational resources
Possible false positives
CONCLUSION
This paper presents a YOLO-based weapon detection and alert system that enhances security by automating
surveillance processes. The system detects weapons in real-time and alerts authorities instantly, reducing
response time and improving safety. Future work can focus on improving accuracy, reducing false positives, and
integrating advanced models like YOLOv8.
Future Work
Integration with IoT devices
Deployment on edge devices
Use of advanced YOLO versions
Improved dataset training
REFERENCES
1. B. Pandey, “Weapon Detection Using YOLO V3 for Smart Surveillance System,” 2021.
2. A. Belurkar, “Weapon Detection using YOLOv4 and CNN,” 2022.
3. R. M., “Real-Time Weapon Detection using YOLOv8,” 2024.
