THE 3
rd
INTERNATIONAL SCIENTIFIC CONFERENCES OF STUDENTS AND YOUNG RESEARCHERS
dedicated to the 99
th
anniversary of the National Leader of Azerbaijan Heydar Aliyev
171
TURNSTILE ACCESS CONTROL BASED ON FACIAL
RECOGNITION AND VACCINE PASSPORT VERIFICATION
Shams Aliyeva
shams.aliyeva.std@bhos.edu.az
Baku Higher Oil School
Baku, Azerbaijan
Supervisor: Ph.D Associate Professor Ali Parsayan
Keywords:
COVID-19, YOLOv3, R-CNN,
Face Recognition
Introduction
Turnstiles are commonly used at different facilities – they allow only one
person at a time to pass through and they manage who is allowed to enter.
Installation of turnstiles allows companies to control entry into their facility
autonomously and efficiently. Turnstiles are smarter and more reliable than
they've ever been due to technological developments over the last decades.
Considering the restrictions due to COVID-19 pandemic,
face recognition
and vaccine passport verification can help touch-free entrance control to be
safer and easier than ever. There are several methods available for object
detection. For instance, a sliding window method
is used by systems like
deformable parts models (DPM), in which the classifier is run at regularly
spaced locations throughout the entire image. Another example can be R-
CNN which uses region proposal method [2]. However,
due to the faster
recognition and performance, object detection is performed with YOLOv3 in
this project. The main goal of this project is controlling access of academic
staff, students and personnel into the university building. They are allowed
to pass through the turnstile only when they have authorization to enter and
valid vaccine passport. Moreover, according to the information obtained from
the database, ones with confirmed COVID-19
will not have an entrance
access.
Algorithm
YOLO is a novel method for detecting and creating bounding boxes
around several objects in a picture in real time. The image is only passed
through the CNN algorithm once to obtain the output, hence the name [1].
YOLO algorithm works using the main three techniques. Residual blocks -
Firstly, the image is divided into various grids. Each grid has a dimension of
S x S. Second step is bounding box regression. A bounding box is an outline
that highlights an object in an image. Every
bounding box in the image
consists of width (bw), height (bh), class (c) and bounding box center (bx,
by). The last step is intersection over union (IOU) which is a phenomenon in
object detection that describes how boxes overlap. YOLO uses IOU to
provide an output box that surrounds the objects perfectly [2].