Real-Time Automated Overspeeding Detection and Identification System
0 likes43 views
This document proposes a real-time automated system to detect overspeeding vehicles using computer vision techniques. The system uses a camera to capture video of vehicles on the road. It detects vehicles using a classifier and estimates their speed using distance and time calculations between frames. For vehicles detected as overspeeding, it performs license plate detection using YOLOv3 and optical character recognition to extract the plate text. This text is then sent to a government database to identify the vehicle owner in order to enforce overspeeding laws. The system aims to automatically detect and identify overspeeding in real-time without expensive equipment or additional traffic police.
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 565
Real-Time Automated Overspeeding Detection and Identification System
Lakshitaa Sehgal1, Anshal Aggarwal2, Sarthak Sood3, Aryan Aggarwal4
1University Institute of Engineering and Technology Chandigarh, Panjab University, Chandigarh, India
2University Institute of Engineering and Technology Chandigarh, Panjab University, Chandigarh, India
3National Institute of Technology, Kurukshetra, India
4Chitkara University, Punjab, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - In today’s world, overspeeding is a big issue and
causes a great number of road accidents. A lot of people get
severely injured and some also lose their lives due to this. So
it becomes essential for traffic police authorities to prevent
it and keep a strict check on people who break permissive
speed limits. So, an automated system is necessary which
detects whether a car is speeding or not. This paper
proposes a real-time automated over speeding detection
and Identification system which checks whether a car is
overspeeding or not using OpenCV and Python. By using
YOLOv3 for object detection, the vehicle number can then be
determined if the vehicle is overspeeding. Then using Optical
Character Recognition (OCR), the text extracted is passed
through the government database system to retrieve
information about the owner.
Key Words: Speed detection, OpenCV, Python, Object
Detection, Person Identification
1. INTRODUCTION
In recent times, where road accidents are increasing day
by day, it has become important to cater to them and keep
a strict check to reduce them immensely. Overspeeding
constitutes approximately 60% of road accidents (as per
2020 reports). Other dangers associated with
overspeeding is that it reduces the effectiveness of airbags,
seat belts and other safety equipment. It can pose a life
threat to pedestrians also.
To check for this, this paper proposes an end-to-end
system using the Cascade Classifier[1] to capture vehicles
on the road and then using OpenCV[2] and Python to
detect their speeds. Using YOLOv3[3], if it is determined
that the vehicle is overspeeding, a license plate detection
is done, which is then converted into text using Optical
Character Recognition (OCR)[4], and this text is passed to
the government database for identification of the vehicle
owner.
2. PROPOSED SYSTEM
This section represents the fully automated and end-to-
end proposed system for overspeeding detection and
Identification.
2.1 SYSTEM SETUP
Fig -1: System Setup
The Raspberry Pi 4[5] and a CCTV camera with NVR for IP
cameras and DVR for analogue cameras or webcam with
attached system memory are used to capture real-time
videos of vehicles on the road. Along with it, an additional
HC-05 Bluetooth Module is used. The hardware system
setting allows for the capture of real-time videos of
vehicles, which are then passed for speed estimation.
2.2 VEHICLE DETECTION
To detect vehicles, the CascadeClassifier function[1] is
used. It's a machine-learning approach in which a cascade
function is trained using a large number of positive and
negative photos. After that, it's used to find items in other
photos. Along with this, a pre-trained XML file
(vech.xml)[6] is used. In this case, multiple objects, such as
cars, need to be identified at the same time, so
detectMultiScale is also used.](https://image.slidesharecdn.com/irjet-v9i294-220913094000-80b2fce9/85/Real-Time-Automated-Overspeeding-Detection-and-Identification-System-1-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 566
Fig -2: Vehicle Detection
2.3. SPEED ESTIMATION AND DETECTION
For speed detection, OpenCV[2] and Python are used. The
basic approach for detecting is to locate two points on the
road and calculate the distance between them. Time is
estimated by calculating the time taken by a vehicle to
cover the distance between those two demarcated points.
Speed is then calculated by applying the distance-time
formula as illustrated below.
Let (x1, y1) and (x2, y2) be the two points where (x₂, y₂)
and (x₁,y₁) represents the current frame’s and previous
frame’s centroid of the vehicle respectively.
Time taken by vehicle to cover distance between two
points = T seconds
Distance in pixels = √[(x₂ - x₁)² + (y₂ - y₁)²]
Speed = Distance/Time
Speed = √[(x₂ - x₁)² + (y₂ - y₁)²] / T
Here, distance is measured in kilometers
Time is measured in hours
So, speed is detected in kilometers/hours
Fig -3: Speed Detection using OpenCV and Python
The above model is tested on some random videos
available online for speed detection and optimum results
have been obtained from this model.
If the vehicle is found overspeeding, then the license plate
of the vehicle is detected.
2.4 LICENSE PLATE OBJECT DETECTION
If a vehicle is found to be overspeeding, an object
detection system records the vehicle's license plate. Here
is YOLOv3 (You Only Look Once, Version 3)[3]. Video, live
feeds, and images can all be analyzed by a real-time object
detection algorithm called YOLOv3[3]. It uses features
learned by a deep convolutional neural network to detect
an object.
The model using YOLOv3[3] for object detection is trained
and tested on the Car Number Plate Detection dataset[7]
which consists of 931 car images and gives the best
accuracy of approximately 85.4%
Fig -3: Before Object Detection
Fig -4: After Object Detection](https://image.slidesharecdn.com/irjet-v9i294-220913094000-80b2fce9/85/Real-Time-Automated-Overspeeding-Detection-and-Identification-System-2-320.jpg)
![International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 567
Fig -5: Before Object Detection
Fig -6: After Object Detection
2.5 OPTICAL CHARACTER RECOGNITION (OCR)
To convert the identified object into text, we use
pytesseract OCR[4] library in python. Tesseract searches
pixels, letters, phrases, and sentences for templates. It
employs a two-step process known as adaptive
recognition. It requires one data step for character
identification, followed by a second stage to fill in any
missing letters with letters that match the word or phrase
context.
Fig -7: Image to Text Conversion
Fig -8: Image to Text Conversion
After the text conversion, the text is passed through the
government database and identification of the car owner
can be identified from the registered car license plate.
3. CONCLUSION
A robust end-to-end Real-Time Automated Overspeeding
Detection and Identification System has been proposed
here which can be used in the future to keep strict on
overspeeding vehicles. This proposed system is more cost-
efficient as no expensive speed guns, heavy speed
detection devices or extra traffic police staff are required
to manually detect speed. This is an end-to-end automated
speed detection and identification system.
REFERENCES
[1] Soo, Sander. “Object detection using Haar-cascade
Classifier.” (2014).
[2] I. Culjak, D. Abram, T. Pribanic, H. Dzapo and M.
Cifrek,”A brief introduction to OpenCV,” 2012
Proceedings of the 35th International Convention
MIPRO, 2012, pp. 1725-1730.
[3] https://arxiv.org/abs/1804.02767
[4] Sargur N. Srihari, Ajay Shekhawat, and Stephen W.
Lam. 2003. Optical character recognition (OCR).
Encyclopedia of Computer Science. John Wiley and
Sons Ltd., GBR, 1326–1333
[5] Maksimovic, Mirjana & Vujovic, Vladimir & Davidovi´c,
Nikola & Milosevic, Vladimir & Perisic, Branko. (2014).
Raspberry Pi as Internet of Things hardware:
Performances and Constraints.
[6] https://docs.cryengine.com/display/CEMANUAL/Veh
icle+XML+Reference
[7] https://www.kaggle.com/elysian01/car-number-
plate-detection](https://image.slidesharecdn.com/irjet-v9i294-220913094000-80b2fce9/85/Real-Time-Automated-Overspeeding-Detection-and-Identification-System-3-320.jpg)
Real-Time Automated Overspeeding Detection and Identification System
- 1. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 565 Real-Time Automated Overspeeding Detection and Identification System Lakshitaa Sehgal1, Anshal Aggarwal2, Sarthak Sood3, Aryan Aggarwal4 1University Institute of Engineering and Technology Chandigarh, Panjab University, Chandigarh, India 2University Institute of Engineering and Technology Chandigarh, Panjab University, Chandigarh, India 3National Institute of Technology, Kurukshetra, India 4Chitkara University, Punjab, India ---------------------------------------------------------------------***---------------------------------------------------------------------- Abstract - In today’s world, overspeeding is a big issue and causes a great number of road accidents. A lot of people get severely injured and some also lose their lives due to this. So it becomes essential for traffic police authorities to prevent it and keep a strict check on people who break permissive speed limits. So, an automated system is necessary which detects whether a car is speeding or not. This paper proposes a real-time automated over speeding detection and Identification system which checks whether a car is overspeeding or not using OpenCV and Python. By using YOLOv3 for object detection, the vehicle number can then be determined if the vehicle is overspeeding. Then using Optical Character Recognition (OCR), the text extracted is passed through the government database system to retrieve information about the owner. Key Words: Speed detection, OpenCV, Python, Object Detection, Person Identification 1. INTRODUCTION In recent times, where road accidents are increasing day by day, it has become important to cater to them and keep a strict check to reduce them immensely. Overspeeding constitutes approximately 60% of road accidents (as per 2020 reports). Other dangers associated with overspeeding is that it reduces the effectiveness of airbags, seat belts and other safety equipment. It can pose a life threat to pedestrians also. To check for this, this paper proposes an end-to-end system using the Cascade Classifier[1] to capture vehicles on the road and then using OpenCV[2] and Python to detect their speeds. Using YOLOv3[3], if it is determined that the vehicle is overspeeding, a license plate detection is done, which is then converted into text using Optical Character Recognition (OCR)[4], and this text is passed to the government database for identification of the vehicle owner. 2. PROPOSED SYSTEM This section represents the fully automated and end-to- end proposed system for overspeeding detection and Identification. 2.1 SYSTEM SETUP Fig -1: System Setup The Raspberry Pi 4[5] and a CCTV camera with NVR for IP cameras and DVR for analogue cameras or webcam with attached system memory are used to capture real-time videos of vehicles on the road. Along with it, an additional HC-05 Bluetooth Module is used. The hardware system setting allows for the capture of real-time videos of vehicles, which are then passed for speed estimation. 2.2 VEHICLE DETECTION To detect vehicles, the CascadeClassifier function[1] is used. It's a machine-learning approach in which a cascade function is trained using a large number of positive and negative photos. After that, it's used to find items in other photos. Along with this, a pre-trained XML file (vech.xml)[6] is used. In this case, multiple objects, such as cars, need to be identified at the same time, so detectMultiScale is also used.
- 2. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 566 Fig -2: Vehicle Detection 2.3. SPEED ESTIMATION AND DETECTION For speed detection, OpenCV[2] and Python are used. The basic approach for detecting is to locate two points on the road and calculate the distance between them. Time is estimated by calculating the time taken by a vehicle to cover the distance between those two demarcated points. Speed is then calculated by applying the distance-time formula as illustrated below. Let (x1, y1) and (x2, y2) be the two points where (x₂, y₂) and (x₁,y₁) represents the current frame’s and previous frame’s centroid of the vehicle respectively. Time taken by vehicle to cover distance between two points = T seconds Distance in pixels = √[(x₂ - x₁)² + (y₂ - y₁)²] Speed = Distance/Time Speed = √[(x₂ - x₁)² + (y₂ - y₁)²] / T Here, distance is measured in kilometers Time is measured in hours So, speed is detected in kilometers/hours Fig -3: Speed Detection using OpenCV and Python The above model is tested on some random videos available online for speed detection and optimum results have been obtained from this model. If the vehicle is found overspeeding, then the license plate of the vehicle is detected. 2.4 LICENSE PLATE OBJECT DETECTION If a vehicle is found to be overspeeding, an object detection system records the vehicle's license plate. Here is YOLOv3 (You Only Look Once, Version 3)[3]. Video, live feeds, and images can all be analyzed by a real-time object detection algorithm called YOLOv3[3]. It uses features learned by a deep convolutional neural network to detect an object. The model using YOLOv3[3] for object detection is trained and tested on the Car Number Plate Detection dataset[7] which consists of 931 car images and gives the best accuracy of approximately 85.4% Fig -3: Before Object Detection Fig -4: After Object Detection
- 3. International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 09 Issue: 02 | Feb 2022 www.irjet.net p-ISSN: 2395-0072 © 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 567 Fig -5: Before Object Detection Fig -6: After Object Detection 2.5 OPTICAL CHARACTER RECOGNITION (OCR) To convert the identified object into text, we use pytesseract OCR[4] library in python. Tesseract searches pixels, letters, phrases, and sentences for templates. It employs a two-step process known as adaptive recognition. It requires one data step for character identification, followed by a second stage to fill in any missing letters with letters that match the word or phrase context. Fig -7: Image to Text Conversion Fig -8: Image to Text Conversion After the text conversion, the text is passed through the government database and identification of the car owner can be identified from the registered car license plate. 3. CONCLUSION A robust end-to-end Real-Time Automated Overspeeding Detection and Identification System has been proposed here which can be used in the future to keep strict on overspeeding vehicles. This proposed system is more cost- efficient as no expensive speed guns, heavy speed detection devices or extra traffic police staff are required to manually detect speed. This is an end-to-end automated speed detection and identification system. REFERENCES [1] Soo, Sander. “Object detection using Haar-cascade Classifier.” (2014). [2] I. Culjak, D. Abram, T. Pribanic, H. Dzapo and M. Cifrek,”A brief introduction to OpenCV,” 2012 Proceedings of the 35th International Convention MIPRO, 2012, pp. 1725-1730. [3] https://arxiv.org/abs/1804.02767 [4] Sargur N. Srihari, Ajay Shekhawat, and Stephen W. Lam. 2003. Optical character recognition (OCR). Encyclopedia of Computer Science. John Wiley and Sons Ltd., GBR, 1326–1333 [5] Maksimovic, Mirjana & Vujovic, Vladimir & Davidovi´c, Nikola & Milosevic, Vladimir & Perisic, Branko. (2014). Raspberry Pi as Internet of Things hardware: Performances and Constraints. [6] https://docs.cryengine.com/display/CEMANUAL/Veh icle+XML+Reference [7] https://www.kaggle.com/elysian01/car-number- plate-detection