Computer vision based smart vehicle security system
Vehicle security is an important issue for all motorists. Every year, around 21,000 cars are stolen and they are worth up to a staggering PKR 10 billion. So, preventing vehicle from theft is an important security problem. Present security system of vehicles depends upon sensors which are sometimes n
2025-06-28 16:30:53 - Adil Khan
Computer vision based smart vehicle security system
Project Area of Specialization Artificial IntelligenceProject SummaryVehicle security is an important issue for all motorists. Every year, around 21,000 cars are stolen and they are worth up to a staggering PKR 10 billion. So, preventing vehicle from theft is an important security problem. Present security system of vehicles depends upon sensors which are sometimes not enough for security in case of theft attempts.
We propose a solution to this problem by controlling access through face recognition and finger print verification to protect vehicle from unauthorized access. Face recognition system is based upon the state of art Residual Neural Networks. The ResNet29 Model which is a variant of ResNet32 is capable of end to end learning providing the accuracy of 99.37 percent on the dataset labelled faces in wild. The finger print provides additional security making the project anti-spoof. The database could be made by the owner himself with ease making it user friendly. The purposed system has a touch enabled LCD display and a user-friendly Graphical User Interface (GUI). With the up-to-date and influential technology, the system is not only expected to be workable, but also sufficiently efficient in terms of execution speed and response time.
The system first detects faces in real environment and then performs face recognition for the authentication of owner, then goes for the finger print verification. If the owner is not authenticated, all the systems of the car remain off. An e-mail will also be sent to owner containing an image of the person who tries to access the vehicle along with current location. We implement this prototype model on an embedded platform Raspberry pi. The Raspberry pi is low cost and controls all the functions of our system. Our embedded solution can also be used for other security applications involving access control.
Project ObjectivesThe objective of this project is to present a model to effectively use state of art facial recognition and fingerprint verification to enhance the vehicle security along with a user-friendly interactive interface. By the hardware/software co-design, the new intelligent deep learning-based vehicle security system implemented the functions of biometric verifications, GPS positioning, user friendly interface and wireless transmission, met the needs of vehicle owners about Vehicle Security.
To accomplish this objective, we sub-divided it into smaller objectives which includes:
- Low cost Hardware Development
- Development of fast and accurate algorithm
- Development of User-Friendly GUI
The aim is to develop a low-cost reliable hardware. The complete design was made on paper considering all the sensors and performance requirements. For this a Raspberry pi based low cost design was considered. Raspberry pi being reliable provides the good performance by consuming the power of 1.2 watts only. Then low-cost Hardware was developed from this design.
Development of Fast and accurate algorithmAs, we decided to use the raspberry pi. There was need for the computationally inexpensive algorithm which also provides the better accuracy. In this regard different algorithm-based approaches were developed and tested for execution time and accuracy. Residual Neural Network based models were selected as they provide much higher accuracy at lower computation expense.
Development of User-Friendly GUIThe user-friendly GUI is important as it makes good device user interaction possible. The GUI will allow to access all the operations in a beautiful via touch enabled LCD. The main hurdle is to link all the algorithm operation with the GUI in a professional and smooth way.
Project Implementation MethodThe main challenges in Computer vision based smart vehicle security system are as:
- Low cost hardware development
- Development of fast and accurate algorithm
- Development of User-Friendly GUI
The low-cost hardware development directed us to use the System on chip (SoC) devices such as raspberry pi 3 model B. Raspberry pi 3 model b offers 1 gigabytes of ram with a 1.2 GHZ quad-core ARM Cortex A53 processor providing enough capability to coup with the less strained processing easily. The following is the hardware integration model of the system.
The overall algorithm for the operation of system is as followed,
As the face recognition is a computationally expensive need for the efficient and faster algorithm.
Face RecognitionThe development of simpler, fast and highly accurate algorithm made us to just end-to-end learning based Resnet29 Model for face recognition. It allows us to achieve an unmatched security level. Model was trained on the GPU and then employed to the raspberry pi 3 model B. The algorithm includes steps which are
TrainingThe faces are mapped onto the 128-dimensional embedding after some preprocessing.
The face to be verified is mapped on to the 128-dimensional embeddings after some preprocessing is checked on the basis of Euclidian distance.
The user-friendly GUI is important as it makes good device user interaction possible. The GUI will allow to access all the operations in a beautiful via touch enabled LCD. The main hurdle is to link all the algorithm operation with the GUI in a professional and smooth way.
Benefits of the ProjectThe major benefits which the project will demonstrate are as mentioned below:
- Protect Vehicles
- Tracks the Vehicle
- Criminal Identification
- Highly accurate
- Deters Crime
- One of the major benefits of this project is that it protects the vehicles from the theft. Controlling access through face recognition and finger print verification to protect vehicle. Using multiple level of verification techniques provides higher degree of security.
- Owner could keep track of the vehicles location and path via GPS modules installed in the system. Moreover, if there are failed attempt to access the vehicle, the owner would be notified with the location of vehicle.
- The owner will be notified with the picture of the person who tries to access the vehicle. This help in the identification of criminal, thus reducing the crime.
- The system is highly accurate as recognition algorithms are based upon the deep learning which incorporates artificial intelligence. Hence, provides surety of correct recognition.
- It can be used in any kind of security system for automatic criminal recognition.
The project delivered would be security system which would have an intelligent recognition system which have ability to
- Ability to enroll more than 3000 person and the figure could be increased by increasing the memory of the system.
- The enrollment of the new person could be done in just 10 sec.
- Works even in Low light using a contrast enhancement algorithm.
- Have the connectivity to the Internet via using GPRS/CDMA.
Following are components used in the system,
Raspberry piThe Raspberry Pi is a series of small single-board computers. It runs on Raspbian which is based on python programming language. Following are specs of raspberry pi
- Broadcom BCM2837 64bit ARMv7 Quad Core Processor powered Single Board Computer running at 1.2GHz
- 1GB RAM
- BCM43143 WiFi on board
- Bluetooth Low Energy (BLE) on board
- 40pin extended GPIO
- 4 x USB 2 ports
- 4 pole Stereo output and Composite video port
- Full size HDMI
The system includes High definition 16 Mega pixel camera providing high resolution picture for our face recognition system @ 30fps.
Gps ModuleThe GPS module used is a highly sensitive module with very precise accuracy. It has ability work on much weaker signals thus making useful in practical scenarios.
Some other components are
- Fingerprint sensor
- Touch LCD
- Power Bank
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 68450 | |||
| Asus Dual GTX 1060 O3G Graphics Cards | Equipment | 1 | 35000 | 35000 |
| Raspberry Pi model B | Equipment | 1 | 7000 | 7000 |
| SD cards class 10 32 GB | Equipment | 2 | 1500 | 3000 |
| Camera Module | Equipment | 2 | 1900 | 3800 |
| Casing | Equipment | 2 | 350 | 700 |
| Touch LCD | Equipment | 1 | 7000 | 7000 |
| HDMI to VGA | Equipment | 1 | 700 | 700 |
| Power Bank | Equipment | 1 | 4500 | 4500 |
| Finger Print Module | Equipment | 1 | 3000 | 3000 |
| Gps Module | Equipment | 1 | 2550 | 2550 |
| Shipping Charges | Miscellaneous | 3 | 400 | 1200 |