weapons and fire detection system
We presented a real-time weapons and fire detection system that assists a human operator by alerting him/her when a potentially firearms or fire is detected. The method uses a YOLOV4 (you only look once) classification model to classify part of a CCTV video frame as either containing a weapons such
2025-06-28 16:30:00 - Adil Khan
weapons and fire detection system
Project Area of Specialization Artificial IntelligenceProject SummaryWe presented a real-time weapons and fire detection system that assists a human operator by alerting him/her when a potentially firearms or fire is detected. The method uses a YOLOV4 (you only look once) classification model to classify part of a CCTV video frame as either containing a weapons such as (Handgun, Knife, Rifle) and events such as fire (small electrical fire, car fire, forest fire, smoke) or not. After detecting it alerts the respected authorities/User by send snap shot of the frame containing that object by the means of e-mail to avoid false alarm and to take action way earlier than expected. This system increases response time as some situation are very time sensitive and require action in initial stage to eliminate it. Eliminates the boundaries of traditions systems such as, human monitoring CCTVs, smoke Detectors, metals detector, security guards etc.
Project ObjectivesAs the number of CCTV cameras installed for surveillance purposes to detect crime is increasing drastically it also becomes practically hectic for human being to be able to detect anomalies and events in environment such as use of weapons for harming someone , or fire eruption which can cause massive damage to environment, So to solve this problem we have come up with a system which automatically detects object such as weapons and events such as fire eruption in real time, which further immediately alerts the user (LEA, Security...Etc.) as these events are time sensitive and require action as soon as possible, to make this possible we use deep learning algorithm and image processing to establish real time object detection through cameras, we also developed a proper application through which user can start detection and stop whenever he need to the application also has a dashboard which is connected to a database, so whenever a detection of objects take place the system takes multiple snaps of that event and alerts user through dashboard which is connected to a database. Through this system we can eliminate the manual monitoring of CCTV cameras, and also real time threats can be eliminated in its initial time.
Project Implementation Methodstep-1 Simulation Model
We created a simulation model of the application by using Adobe XD and designing GUI of the application Figma
Step-2 SDLC
In this phase we planned the execution of the project, from deigning, implementation of code by using python programming language, Test the final prototype in real world scenario, and deploying the project.
Step-3 Functionality
1-sign in and registration form
2-selecting detection type
3-luanching of the webcam and detection
Step-4 Integration
In the integration we connected the hardware such as graphics card, web cam to the application
Step-5 Testing
We tested the projefct in real life situtation, tested using real life object such as knife, gun. fire etc.
Benefits of the Project1. Establishing a camera based automatic fire and weapon recognition system.
2. Which can achieve 24/7 automatic monitoring, which will greatly reduce labor cost
3. Establishing system in environment where it is needed i.e. schools, university, hospitals, offices etc.
4. Alerting administration, fire brigade, police to stop fire and use of weapons such as guns, as these cases are time sensitive.
5. Eliminating traditional methods such as manual cameras monitoring and smoke detectors which only alerts user when fire is already spreded.
6. Minimizing false alarm by using better deep learning algorithm such as CNN (convolution neural networks)
Technical Details of Final Deliverable- Programming Languge (Python 3.9)
- Anaconda (jupyter Notebook)
- Python Libraries( Numpy, SMTPlib, TKinter, OpenCV-CV2)
- Deep learning Model (YoloV4)
- Labeling tool (labelimg)
- Designing Tool (Adobe XD, Figma)
- Graphics Card (Geforce 750Ti)
- Operating System (Windoes 10)
- RAM(8GB)
- Processor(intel i5 4th generation)
- WEB-Cam (logitect 720HD)
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 46050 | |||
| GPU | Equipment | 1 | 30000 | 30000 |
| Web-Cam(sensor) | Equipment | 1 | 4000 | 4000 |
| RAM(8GB) | Equipment | 1 | 2500 | 2500 |
| USB | Equipment | 1 | 650 | 650 |
| project report printing | Miscellaneous | 1 | 5100 | 5100 |
| FYP overheads, | Miscellaneous | 1 | 3500 | 3500 |
| Stationery(Files,Markers) | Miscellaneous | 0 | 1000 | 0 |
| CDs | Equipment | 3 | 100 | 300 |