2022 8th International Conference on Smart Structures and Systems (ICSSS)

Cloud Computing Based Face Mask and Helmet Facial

Detection for ATM Security Using Image Processing
B.Veena S.Babu
Department of Computing Technologies, Department of Computing Technologies,
SRM Institute of Science and Technology Kattankulathur SRM Institute of Science and Technology Kattankulathur
Tamil Nadu, India Tamil Nadu, India
[email protected] [email protected]

Abstract — In this modern-day scenario the ATM robbery are and it can be embedded device.The total process is defined as
increasing rapidly thus the security is one of the biggest SSDMNV2.This technique is used to analyse the accuracy of 0.9264
concerns and this project will focus on improving the security of and F1 score is 0.93.[2]
2022 8th International Conference on Smart Structures and Systems (ICSSS) | 978-1-6654-9761-9/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICSSS54381.2022.9782202

ATM transactions. The main process of the project would be

incorporating cloud and image processing. Image processing The model which is subdivided into two phases, where in
will play a crucial role in the current situation where we will first phase the face is detected except the cases which the object is
need data to be processed for differentiating between people kept far away and second phase where the face is detected as a
who are wearing mask and not wearing mask. Though every individual. The process is repeated by helping in recognizing the
people have to wear mask as a safety precautions, some people face model which is nowadays used for biometric technology [3].
misuse it for illegal things. The project’s current focus is only on Showed some of the recent image processing applications in the
ATM transactions. In this project the process of ATM cloud environment which are reviewed and discussed. Some popular
transaction will be finetuned so that a person can’t proceed in image processing tools which are working in the cloud environment
the transaction without removing the mask and helmet. With are also reviewed [4]. Image processing classification on small-scale
each transaction the image of the person along with the with limited number of hardware on which we are possession of. As
transaction detail will be stored in cloud from which we can in this project where SVM is used to provide high accuracy of 93%.
securely access anywhere and anytime (with proper SVM is very strong technique by using this technique high accuracy
authentication). can be achieved [5].
Keywords — Mask, Helmet, Image Processing, Cloud and
Transactions A system where it will help in capturing the real time human
I. INTRODUCTION (HEADING 1) face. This face detection implementation is done in smart phone,
According to today’s world people are getting effected by machines and many other software applications [6]. A face detection
climate change, infectious disease and pollution. To overcome these system for the utilization of computer learning, especially Open CV.
problems, it became as a basic need for a person to wear mask in The mandatory step required is face detection which we did with the
public area. Sometimes situation is taken as an advantage by thief's usage of a broadly used step referred to as the haar-cascade frontal
and intruders to do their illegal activities without showing their face face default classifier, python and its module [7].
in CCTVs and cameras. Many people take money from ATM
without revealing their faces. These kinds of activities create a A face detection technology based on computer vision-
misunderstanding between customer and bank. To overcome this based algorithm. Seetaface method and YouTu method is used to
issue, It is necessary to revel the face of the person in bank ATM improve the face recognition technology. On the same time the
camera to do their transaction via ATM. This will be achieved by experiment of detection and recognition is done under three different
image processing where the system will be able to find the requirements that is occlusion detection, face detection and facial
difference between person wearing a mask or not wearing. If any exaggerated which are compared and accuracy of each is improved
person performing transaction by wearing mask then the transaction [8].
will not take place and displays a warning message to remove mask
and helmet then only the transaction can take place further. The They examine the question, which involves the thought of
large data which we collect daily can be stored in virtual database face recognition using incomplete facial information. The experiment
that is cloud. Stored data can be used for future references. is predicated on the utilization of Object-Oriented programming
language (OOP) with OPENCV (Open Computer Vision) for correct
II. RELATED WORK classification and identification of the face[9].To detect identifying
the face in downside and it is concatenated to one feature which is
For detecting face mask on people where they use a employed to coach the system to recognize the static images. This
method like TensorFlow and Open CV. When the face is detected process is done in trained system where classification is initiated into
the bounding box will be formed around the face which detects it. The algorithm used to recognize the face image and acknowledges
weather the person is wearing mask or not. If the person’s face is it [10].
stored in database, then the name person who are not wearing mask
will be detected and email will be sent as a warning message[1].A
face detection method where they use Single shot multibox detector
and to detect real-time mask MobileV2 is used which is lightweight
978-1-6654-9761-9/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: PES University Bengaluru. Downloaded on December 11,2022 at 07:14:36 UTC from IEEE Xplore. Restrictions apply.
 2022 8th International Conference on Smart Structures and Systems (ICSSS)

III. SYSTEM ARCHITECTURE taking place, so based upon the comparison the decision is made on
the processing.

Fig.3. Real time image flow diagram

The embedded camera is used to detect the real time images after
it gets detected the preprocessing on the dataset is done and using
deep algorithm method that is using MobileNetV2 the algorithm is
applied and it is converted into video frame where it displays the face
within the boundary with accuracy.

V. PROPOSED WORK

The main aim of my project is to have authorized and secure

transaction processing via ATM So if we have smooth transaction
then no customer will be cheated and there won’t be any
misunderstanding between the bank and customers .And if we have a
correct count of the customer and exact person who are doing
transaction then if any problem occurs in future then these will be
used as a references to identify the problem.

Fig .1. System Architecture VI. METHODOLOGY

IV. DATA FLOW DIAGRAM A. Processing of static image

As this module is based upon analyzing the stored static
image and process the data using deep learning algorithm. In
this when the face recognizance using deep learning is done
where greater accuracy can be achieved when compared to
traditional machine learning methods. The process is done in the
form of Convolution Neural Network (CNN’s) using this
method in deep learning facial recognition is done. The CNN
contains multiple layers that is called neurons where these
neurons consist of input layer, output layer and multiple layer
in-between. In the input layer where the images are scanned in
the form of matrices which is further multiplied the result is fed
into multiplication layer. The process continues till the output
layer is generated in the form of array.

B. Processing of real time image

In this module where the real time images are captured and
checked weather, they are wearing mask/helmet or not. So, the
deep learning models are developed using python, keras and
OpenCV, dnn (deep neural network) so that using this it is
possible to capture the real time process. To do that process the
Fig.2. Static image flow diagram model is first trained and after training the model is tested using
the embedded web cam. The input for this process is real time
The dataset with mask and helmet, without mask and helmet are human being wearing a mask/helmet or not and the output
added has a input where further feature engineering is done with the where it contains the top-left corner coordinates, height and
help of preprocessing method and comparison of each dataset is width of the rectangle encompassing the faces.
978-1-6654-9761-9/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: PES University Bengaluru. Downloaded on December 11,2022 at 07:14:36 UTC from IEEE Xplore. Restrictions apply.
 2022 8th International Conference on Smart Structures and Systems (ICSSS)

VII. RESULT AND DISCUSSIONS

The precision of the dataset where the ability of the

classification which identifies relevant data points with mask
and helmet is 97% and without mask and helmet is 99%. The
accuracy of a classifier which gives total correct prediction
based upon the dataset is 98%.

Face mask and helmet detector training accuracy and loss

curves demonstrate high accuracy and little signs of overfitting on
the data, as we can see 98% of accuracy is obtained.

Fig.6.without mask

Fig.4.Accuracy Detection

Fig.7.With Mask

VIII. CONCLUSION

The main aim of this project is to have authorized and

secure transaction processing via ATM. Because some illegal
transaction occurring in ATM by wearing a mask and helmet which
leads to customers getting cheated and the misunderstanding occurs
between the bank and customers. So if any problem occurs they
cannot able to identify the particular person therefore to avoid these
Fig.5. graph based of training loss and accuracy
problem only the face mask facial detection using image processing
brought up. So we can have smooth transaction then no ATM will
The real time images are captured and using the embedded be robbed.
camera photos are taken to check whether the person with and
without mask and helmet for further process. REFERENCES
[1]. Harish Adusumalli, D. Kalyani, R.Krishna Sri,
M.Pratapteja, “Face Mask Detection Using OpenCV ” PVR
D Prasada Rao, 978-0-7381-1183-4/21/$31.00 ©2021 IEEE 1309

Authorized licensed use limited to: IEEE Xplore. Downloaded on

September 21,2021 at 14:10:53 UTC from IEEE Xplore.,2021.

[2]. Preeti Nagrath , Rachna Jain , Agam Madan , Rohan

Arora , Piyush Kataria , Jude Hemanth “SSDMNV2: A real time

978-1-6654-9761-9/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: PES University Bengaluru. Downloaded on December 11,2022 at 07:14:36 UTC from IEEE Xplore. Restrictions apply.
 2022 8th International Conference on Smart Structures and Systems (ICSSS)

DNN-based face mask detection system using single shot multibox

detector and MobileNetV2” 2210-6707/© 2020 Elsevier Ltd.

[3]. Gurlove Singh, Amit Kumar Goel , “Face Detection

and Recognition System using Digital Image Processing”,
10.1109/ICIMIA48430.2020.9074838, 23 April 2020.

[4]. Neeraj Kumar Pandey, Manoj Diwakar, “A Review on

Cloud based Image Processing Services”, Conference: 2020 7th
International Conference on Computing for Sustainable Global
Development (INDIACom).

[5]. Sai Yeshwanth Chaganti, Ipseeta Nanda, Koteswara

Rao Pandi, Tavva GNRSN Prudhvith, Niraj Kumar, “Image
Classification using SVM and CNN”,
10.1109/ICCSEA49143.2020.9132851, 2020 International
Conference on Computer Science, Engineering and Applications
(ICCSEA).

[6]. Tejashree Dhawle,Urvashi Ukey, Rakshandha

Choudante , “Face Detection and Recognition using OpenCV and
Python”, Volume: 07 Issue: 10 | Oct 2020.

[7]. Dr. Priya Swaminarayan, Mr. Manoj Nath, Miss.

Bipasha Mandal, “Face Detection with Machine Learning and Open
CV Classifier” , ISSN: 2456-5660 Volume 5, Issue 6, Nov-
December 2020.

[8]. Di Lu, and Limin Yan, “Face Detection and

Recognition Algorithm in Digital Image Based on Computer Vision
Sensor”, Volume 2021 |Article ID 4796768 |13 Sep 2021.

[9]. Jeevan Singh, Vishal Kumar, “Face Detection and

Tracking Using OpenCV: A Survey”, Volume 8, Issue 6 June 20,
IJCRT2006237.

[10]. J. Manikandan, S. Lakshmi Prathyusha, P. Sai

Kumar, Y. Jaya Chandra, M. Umaditya Hanuman, “Face Detection
and Recognition using Open CV Based on Fisher Faces Algorithm”,
Volume-8 Issue-5, January 2020

978-1-6654-9761-9/22/$31.00 ©2022 IEEE

Authorized licensed use limited to: PES University Bengaluru. Downloaded on December 11,2022 at 07:14:36 UTC from IEEE Xplore. Restrictions apply.