Vehicle Recognition System
Summary With the advancements in automobile industry, the automobile architecture has also been evolved greatly. Many imported vehicles are getting popular in Pakistan. The project is basically a system that is able to identify a vehicle type, its manufacturer, generation and ye
2025-06-28 16:36:34 - Adil Khan
Vehicle Recognition System
Project Area of Specialization Artificial IntelligenceProject SummarySummary
With the advancements in automobile industry, the automobile architecture has also been evolved greatly. Many imported vehicles are getting popular in Pakistan. The project is basically a system that is able to identify a vehicle type, its manufacturer, generation and year of the vehicle. As in Pakistan such kind of work to identify vehicles through just images has not been done yet hence the dataset for local vehicles is to be collected by our self. Later on the system will be trained for local vehicles. New models of cars are released every year system will classify them on the bases of vehicle type and its generation. In past recent years inaccurate weight measurement of overloaded vehicles lead to destruction of roads and road accidents. To overcome this, the weight of passing by vehicle is calculated accurately to make sure whether the vehicle conforms to the travelling regulations of safety, load, emissions etc. A person with no or a small knowledge of vehicles will be able to identify the vehicle characteristics by the help of this system. He can inspect the vehicle for various purposes. He can inspect the vehicle for various purposes.
Objectives
Pakistan being under develop country is still not aware of much technology that is in use in other countries. We are aimed to bring these researches to Pakistan and to implement the system in practical applications in Pakistan. There is no local dataset available for local available vehicle of Pakistan. We will be developing our own local dataset to train system on. This desktop application will be able to detect the vehicle from a captured image along with other details about vehicle manufacturer, generation, year of generation, plate number (optional) etc. Our objective is to make a system that the utilizer opts to utilize for getting the information cognate to the vehicle's model and many other cognate features of the car after the image is passed through system which will process the image and reveal its required features being specified by system. Moreover, we are trying to make a system that can be incorporated with the city spot system for better security purposes as the city spot’s system is able to get desired information only when the number plate is visible in the image, this systems should be able to process the image even of the number plate is unreadable and recognize the vehicle so it will take the city spot’s system to next enhanced level.
- Research existing solutions
- Requirement Gathering
- Learning new technologies
- Designing architectures
- Functional Requirements
- Use case Diagrams
- Entity Relationship Diagram
- Sequence Diagram
- Activity Diagram
- Literature Review
- Report Submission
- Dataset Building Transfer Learning
- Model Training Phase
- Model Testing Phase
- Application Development
- Interface Front end
- Interface Back end
- Integrating Front and Back ends
- Debugging Testing
- Deployment
Future Work & Benefis
Being an important branch of ITS, image based fine-grained motor vehicle recognition has become the focus of many researchers in past years. Despite of that the works have been done in this regard and a lot more is yet to come unfortunately Pakistan being under develop country is still not aware of this technology. By getting the information from these results of fine grain vehicle recognition and weight system, an intelligent system can be developed that could monitor the whole city roads. Further analyses of the data can help the authorities to monitor the behaviour of vehicles to find out the routines of citizens which ultimately enables to schedule transportation rules accordingly in real time to develop a smart city where tracking, search, re-identification of vehicles becomes much easier. Location access could be an important feature to add in future.
Since the problem is a classification problem, machine Learning Classification techniques will be used to train the model. The most suitable algorithm of machine learning would learn over to our created data set of Pakistani cars. The system will be trained to recognize the Vehicle from the image of the vehicle, provided the image is from frontal angles. Recognizing includes the following
- Manufacturer of vehicle
- Generation of the vehicle
- Year of the generation
- Number Plate ( optional )
There will be an android application. The image of the car may be taken from the phone camera or the image from other sources. Desktop application and IOS application would also to be work on later.
Final Deliverable of the Project Software SystemCore Industry TransportationOther Industries IT , Others , Security Core Technology Artificial Intelligence(AI)Other Technologies Others, Big DataSustainable Development Goals Industry, Innovation and Infrastructure, Sustainable Cities and CommunitiesRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 80000 | |||
| Cameras, sensors | Equipment | 2 | 10000 | 20000 |
| Graphic card MSI GeForce GTX 1660 Ti VENTUS XS 6G OC | Equipment | 1 | 50000 | 50000 |
| Stationary | Miscellaneous | 1 | 5000 | 5000 |
| Printing | Miscellaneous | 1 | 5000 | 5000 |