CSEIT206276 | Accepted : 01 April 2020 | Published : 07 April 2020 | March-April-2020 [ 6 (2) : 241-246 ]
International Journal of Scientific Research in Computer Science, Engineering and Information Technology
© 2020 IJSRCSEIT | Volume 6 | Issue 2 | ISSN : 2456-3307
DOI : https://doi.org/10.32628/CSEIT206276
241
Cost effective Parking System Using Computer Vision
Kaushal Shah, Shivang Rajbhoi, Nikhil Prasad, Charmi Patel, Roushan Raj
Computer Science and Engineering, Parul Institute of Technology, Vadodara, Gujarat, India
ABSTRACT
This paper presents an approach for detecting real-time parking slots which includes vision-based techniques.
Traditional sensor-based systems are not cost effective as 'n' number of sensors are required for 'n' parking slots.
Transmitting sensor data to central system is done by hardwiring or installing dedicated wireless system which
is again costly. Our technique will overcome this problem by using camera instead of number of sensors which
is expensive. For detection we are using a Convolutional Neural Networks (CNN) classifier which is custom
trained. It is more robust and effective in changing light conditions and weather. The following system do not
require high processing as detections are done on static images not on video stream. We have also demonstrated
real-time parking scenario by constructing a small prototype which shows practical implementation of our system.
Keywords : Convolutional Neural Networks, You Only Look Once, Deep learning.
I. INTRODUCTION
Since last 10 years, there is a huge increase in number
of vehicles. But the current transportation
infrastructure and car park facilities are insufficient in
sustaining the influx of vehicles on the road. Therefore,
problems such as traffic congestion and insufficient
parking space inevitably crops up [1]. Car parking
occupancy detection are of great importance for an
effective management of car parking lots. Knowing in
real-time availability of free parking spaces and
communicating to the users can be of great help in
reducing the queues, improve scalability and the time
required to find an empty space in a parking lot [2]. An
important requirement for these systems is the ability
to detect nearby parking slots automatically, which is
becoming an increasingly difficult and critical task and
therefore the traffic increases [3]. Various measures
have been taken in the attempts to overcome the
traffic problems. Some methods like sensor-based and
computer vision classifiers are functional nowadays to
reduce the problems regarding parking. But these
systems are either very costly or less efficient and less
robust. Current detection systems repurpose classifiers
to perform detection. To detect an object, these
systems take a classifier for that object and evaluate it
at various locations and scales in a test image. Systems
like deformable parts models (DPM) use a sliding
window approach where the classifier is run at evenly
spaced locations over the entire image [4].
We are using YOLO (You Only Look Once), which is
an object detection algorithm based on Convolutional
Neural Network (CNN). YOLO is refreshingly fast as
well as simple and YOLO trains on full images and
directly optimizes detection performance.
Since we frame detection as a regression problem, we
don’t need a complex pipeline. We simply run our
neural network on a new image at test time to predict
detections [5]. YOLO reasons globally about the image
when making predictions. Unlike sliding window and
region proposal-based techniques, YOLO sees the
entire image during training and test time so it