AI BASED THUNDERCAM USING RASPBERRY PI AND IOT
Our proposed system of CCTV Camera using Raspberry PI and IoT is named as ThunderCam. ThunderCam uses its trained dataset for Traffic Rule violation, Robbery Detection and motion detection?s. Real-time human actions identification systems are very important for security, surveillance, and also
2025-06-28 16:25:03 - Adil Khan
AI BASED THUNDERCAM USING RASPBERRY PI AND IOT
Project Area of Specialization Artificial IntelligenceProject SummaryOur proposed system of CCTV Camera using Raspberry PI and IoT is named as ThunderCam. ThunderCam uses its trained dataset for Traffic Rule violation, Robbery Detection and motion detection’s. Real-time human actions identification systems are very important for security, surveillance, and also for a number of other applications. There are numerous approaches to identify, track and recognize individuals in open territories, for example, air terminals, shopping malls, and in zones with limited access, for example, private workplaces, houses, schools etc.
The average cost of a Smart CCTV Camera used in Smart City Projects lays around 70,000-120,000 Pakistani Rupees as mentioned above in which 40 percent of the amount is in terms of taxes of various forms like: Customs Duty, Sales Tax, Income tax and Additional taxes.
Nowadays increase in traffic and violations in its rules are taking place due to which several incidents are happening, we can overcome that by catching the offender this can take place when we have an appropriate automation system for the notification of the lawbreakers. Furthermore, this System consist of a website that will generate notifications when detection is countered by the ThunderCam. In this course of action our System provides a facility for detecting and sending notifications at the very moment.
The heavy taxation that are implied on importing Smart AI CCTV Cameras imposes a massive burden on our nation’s economy. The main goal is to manufacture a Pakistan build Smart Economic Camera that will give an adequate competition to imported Cameras. The local manufactured Camera will subsequently have a low-cost as compared to those being imported.
Project ObjectivesOn the comprehensive standpoint, the goal of the project is to create such system that will not only be economic to any common man but will also provide all the necessities stakeholders in Pakistan looking for. Main objectives of the Project are set down as follow:
- Setting up the respective Raspberry Pi Environment.
- Streaming the Live feed getting from CCTV to Raspberry PI.
- Utilizing Open CV/TensorFlow for Training and availing the Data-sets of Traffic Rules Violation and Robbery Detection.
- Utilizing Open CV/TensorFlow for Training and utilizing the Data-sets for Robbery Detection.
- Making use of already installed CCTV Cameras by connecting them to this system using Raspberry PI.
Interfacing of CCTV Camera with Raspberry Pi:
To interface a network CCTV Camera, we first settled up its environment using SADP Tool which included setting up its IP Address, Password, and other network necessities. After that we tested the CCTV Camera via HIKVISION Website. This was done to ensure that there was no error in our Network Camera after that we started the step to get the Live-stream via Raspberry Pi by installing multiple packages and dnsmasq named DNS server to assign a specific IP to the ethernet port each time the camera is connected to it.
Starting with Tensor Flow & Open CV:
Now after the testing step, we can will shift towards the step of training the data-sets of Traffic Rule Violation and Robbery Detection. This is done using Machine Learning Algorithm i.e: Tensor Flow and deployment is done using Open CV/TensorFlow Lite.
Deep artificial neural networks known as convolutional neural networks are commonly utilized in picture classification, image grouping, and object detection within scenes.
A convolution is the integral measure of how much two functions overlap when one passes over the other in mathematics. Each of the convolutional nets gathers distinct signals since they pass numerous filters over the picture. One may visualize the convolution net crossing a horizontal line, a vertical line, and a diagonal line at a very early stage, forming a map of the image's edges.
Methodology:
The heavy taxation that are implied on importing Smart AI CCTV Cameras imposes a massive burden on our nation’s economy. The main goal is to manufacture a Pakistan build Smart Economic Camera with the Pakistan based scenario that will give an adequate competition to imported Cameras and work more efficiently according to the environment of the homeland. The local manufactured Camera will subsequently have a low-cost as compared to those being imported. This cost can further be reduced and efficiency will be improved once the FPGA Board is designed for the system that we have proposed. At this stage this scenario falls outside the Scope of the Project.
A Smart AI based surveillance system for traffic Rule violation and Robbery Detection. A system that is within the reach of every common man.
Technical Details of Final DeliverableThe ThunderCam when fully interfaced with all its hardware will be able to make Smart Decisions thus reducing the human interactions and involvement as less as possible. Human force will be needed whenever the notification upon detection of a specific scenario happens and thus notification is generated on the respective website alerting the particular authorities.
For the purpose of Training the data of both cases i.e: (Traffic Rule Violation & Robbery Detection) we are using TensorFlow and after the testing deployment is done using TensorFlow Object Detection Api & Open CV/TensorFlow Lite.
The main aim to use TensorFlow is because TensorFlow bundles together Deep Learning and Machine Learning Algorithms and Models. TensorFlow provides a sophisticated framework for Training and deploying models using Python or JavaScript as desired. Furthermore, we will be using TensorFlow Object Detection Api as it open-source framework that makes it easy to construct, train, and deploy models.
In this Api of TensorFlow a large set of bounding boxes is generated at the Image after which visual. After those necessary visual features are extracted from the bounding boxes. They are evaluated and upon evaluation it is decided whether which objects are present in those boxes based upon Feature Classifier technique. And lastly overlapping boxes are overlapped using non maximum suppression method. This Api of TensorFlow has a pre-trained models which are referred as Model Zoo.
The collected dataset is by the team is the data of the Pakistan’s environment that makes the trained model work more efficiently in the environment of Pakistan as the model will be more accurate to detect whether the law has been breaked or not.
Moreover, along with reducing the overall cost of per unit we focused on the main motto of a surveillance camera that is to provide security. In this system we are commencing with two types of Crime Detection, i.e: Traffic Rule Violation and Robbery Detection. This design will measure the parameters at a time with modern technology and using lower power consumption technology. Our design of ThunderCam is such that it is low-cost, low power consuming Smart Monitoring device.
The approach to connect the network CCTV camera to raspberry pi and processing the stream got from it basically focuses on utilizing the installed and functioning normal network cameras across the country by just connecting the to the AI equipped raspberry pi board. It will prevent the installed camera of being electronic waste and also prevent the procurement of further camera to meet the hardware requirements of the developed system.
Final Deliverable of the Project HW/SW integrated systemCore Industry ITOther Industries Manufacturing , Others , Telecommunication Core Technology Artificial Intelligence(AI)Other Technologies Internet of Things (IoT), Shared EconomySustainable Development Goals Decent Work and Economic Growth, Industry, Innovation and Infrastructure, Sustainable Cities and Communities, Peace and Justice Strong InstitutionsRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 76150 | |||
| Raspberry Pi Model 4B | Equipment | 1 | 38500 | 38500 |
| Hikvision IP Camera | Equipment | 1 | 4500 | 4500 |
| 7 Inch HDMI Capacitive Touch LCD Screen | Equipment | 1 | 9900 | 9900 |
| Ethernet Cables | Equipment | 2 | 950 | 1900 |
| SD Card | Equipment | 1 | 950 | 950 |
| Cables | Equipment | 2 | 700 | 1400 |
| Coral USB Accelerator TPU Coprocessor | Equipment | 1 | 10500 | 10500 |
| Led Monitor | Miscellaneous | 1 | 8500 | 8500 |