Automatic-Number-Plate-Recognition

Srishti 2020

ABSTRACT

Automatic Numper Plate Recognition system is a python based utility that captures live traffic, detects number plate of the oncoming vehicles and stores the data into a database.

MOTIVATION

In this fast growing world, crime is increasing exponentially and that calls for better security and survellience in many sensitive and crowded areas like Colleges, Malls etc. Before the emergence of AI, record keeping had to be done manually which led to many human errors, and eventually led to security lapses. With the resources now at our hands, programs like ANPR become a necesity for the greater good of society.

So we took the project to design the above mentioned system, right from the source code to the body of this system.

Workflow

Mechanical Aspects of Design

• Structure A box with 3 sliding doors, to hold the electronic components of the system. Easy to fabricate design, without the hassle of drilling, tapering and milling. Extruded cut in the front to fit the face of camera.

• DIMENSIONS: BOX : 17 x 11 x 10 (cm)

• Model

Electric Aspects of Design

• Microcontroller Raspberry Pi 3B+

• Camera LOGITECH WEBCAM

• Battery 12 volt battery powering the pi with Micro USB 2.0 outlet

Cost Structure

Components	Cost(INR.)
Raspberry Pi	3830.00
Battery	1500.00
LOGITECH WEBCAM	3500.00
Total	8830.00

Applications

• Shopping Malls : To keep a record of vehicles entering
• Toll Plaza : On highways for security reasons
• Parking Plazas : To detect and generate tax for parked vehicles automatically
• Educational or Government Institutions : To ensure only authorized vehicles are permitted inside

Limitations

• Restrictions on Camera's Field Of View , Resolution an Frame Rate .
• A better OCR trained on strong GPU's will give better results .

Future Improvement

• Developments in ML and DL can give more precise and accurate outputs .
• Reduction in size of camera and microprocessor can save space .

Team Members

1. Atharva Karanjgaokar
2. Kushagra Babbar

Mentors

1. Rishika Chandra

NOTE

DOWNLOAD YOLO WEIGHTS FROM HERE

Refernces

• Opencv - https://www.youtube.com/playlist?list=PLvVx8lH-gGeC8XmmrsG855usswhwt5Tr1

• Yolo object dtection - https://pjreddie.com/darknet/yolo/

• Pytessercat - https://www.pyimagesearch.com/2017/07/10/using-tesseract-ocr-python/

• Object tracking - https://www.learnopencv.com/object-tracking-using-opencv-cpp-python/