Finding Missing Person Using Artificial Intelligence

A.Ponmalar1 B.Sandhiya2 M.Bhuvaneswari 3

Department of Information Department of Information Department of Information
Technology Technology Technology
Sri Sai Ram Institute of Technology Easwari Engineering College Sri Sai Ram Institute of Technology
Chennai, Tamil Nadu, India ,Chennai, Tamil Nadu, India Chennai, Tamil Nadu, India
[email protected] [email protected] [email protected]
2022 International Conference on Computer, Power and Communications (ICCPC) | 979-8-3503-9784-0/22/$31.00 ©2022 IEEE | DOI: 10.1109/ICCPC55978.2022.10072122

M.Gayathri 4 GKR Bhavana5 Aarthi S6

Department of Information Department of Information Department of Information
Technology Technology Technology
Sri Sai Ram Institute of Technology, Sri Sai Ram Institute of Technology, Sri Sai Ram Institute of Technology,
Chennai, Tamil Nadu, India Chennai, Tamil Nadu, India Chennai, Tamil Nadu, India
[email protected] [email protected] [email protected]

Abstract—Finding a missing person these days is an KNN classifier technique. We considered our possibilities
extremely difficult effort; even if social media keeps us and settled on this algorithm because of its capacity for
all informed, it still necessitates a lot of paperwork, analysis. As a result of the decreases in expenditures for
takes a lot of time, and offers little guarantees that the labour, capital, and time, the work at hand becomes simpler.
search will be successful. This project provides a system Missing children or any other person is a serious problem
that streamlines facial recognition searches for the that needs to be addressed by society. We made the decision
benefit of the public and the police department. In order to create a system that will use machine learning, deep
for this system to function, the appropriate guardian of learning, and artificial intelligence to help locate missing
the missing individual must upload the image, which will people by recognising their faces. Finding the missing
then be stored in our database. The next step is for our individual is one of the numerous uses for face recognition
system's face recognition algorithm to locate a match for technology, and it has several benefits overall.
that person in the database. If a match is discovered, the To make the work of locating the missing person
police and that person's guardian of that will be alerted. simpler, we intend to create an application that will be used
Face recognition is a biometric technology that creates a by a few volunteers and allow us to quickly locate the
face print by mathematically mapping a certain person's missing person. This will make it simpler for police to find
or individual's facial traits. Using this method, a a certain person. In the meanwhile, automation is required
person's facial information is mathematically or in the to automate the process of locating a certain person by
form of graphs kept in the database that is used to identifying a specific image and comparing that image with
identify that specific face. Our system's face recognition another image to determine whether or not both
model will locate a match for that person in the photographs share the same qualitiesBy doing this, we can
database. The police and the child's guardian will be determine whether the missing person in an image taken
alerted if a match is discovered. from a certain location is real or fake, and if it is, the police
can take further action to locate the individual from that
spot. As a way for the system to recognise that image data
Keywords—Face recognition, Missing person, Recognition. and find the missing individual, our programme will have
the capability of saving all the data of the missing person.

I. INTRODUCTION II. RELATED WORKS

The project's main objective is to find missing
This project's goal, put simply, is to assist in case persons. Along with Postgres Database, we are using Image
investigations and the prompt identification of victims. Processing and Machine Learning[1]. Face recognition is a
Time is of the importance thus. In this case, we'll apply the biometric-based technique that creates a face print by

979-8-3503-9784-0/22/$31.00 2022
c IEEE 562

Authorized licensed use limited to: Zhejiang University. Downloaded on October 24,2024 at 14:08:00 UTC from IEEE Xplore. Restrictions apply.
mathematically mapping a certain person's or individual's
facial traits[4]. Using this method, a person's facial
information is mathematically or in the form of graphs kept
in the database that is used to identify that specific face[8].
Our system's face recognition model will locate a match for
that person in the database. If a match is discovered, the
police and that person's guardian will be alerted[10].

We conducted extensive research and came to the

following conclusions on the literature review. To start, S.
AYYAPPAN and his engineering classmates from the IFET
College of Engineering delivered a work that addresses a
problem statement and purpose that are comparable to our
own. They use a layered convolutional auto encoder and
facial feature extraction and matching based on deep
learning in their suggested solution (SCAE). A database
contains the photos of the missing people. From those
photographs, faces are recognised, and a convolutional Figure 1:Architecture Diagram
neural network picks up features. To appropriately identify
and label the child, a multi-class SVM classifier was trained Numerous datasets are taken into consideration and
using these newly learned features[12]. Shefali Patil and his used in the data analysis. Beginning with a primary focus on
colleagues from the SNDT Women's University in Juhu, the video used to gather locations and faces, as well as the
Mumbai, have already published a study that addresses a face recognition system's knowledge of facial movements
similar problem statement and purpose.They use the KNN and person descriptions, this article's work begins with these
Algorithm, which requires 136 * 3 data points to detect topics. Initial recognition of a person's face edges is the
faces, in their suggested system. The KNN method's most important step in the Robert edge detector method.
accuracy, which is 71.28%, is its biggest drawback[13]. The Then a series of arithmetic operations between the starting
study paper from the Late G.N. Sapkal College of frame and the closest frames follows.
Engineering, which likewise dealt with a related problem
description and purpose, was given by Birari Hetal and her
classmates. The challenge of finding a missing individual
has been made simpler with the creation of an Android
app[14,15]. They employ SWFSIFT as the method for
comparing two photos in the Android application they have
presented[16,17].

III. METHODS AND MATERIALS Figure 2:Process Flow

Testing using actual data, rather than on rules derived later,
Includes a number of elements, such as a dataset with
is the most effective way to evaluate an algorithm's
examples of people who have gone missing and have been
predictive power. A common technique for calculating the
found alive or dead, the latitude and longitude of that
learning scheme's error rate on the obtained data is cross-
specific site, the age of the person, the reason they were
validation.A case is opened with a few images and details
reported missing, and potential locations (latitude and
like name, age, complexion, etc. for every person who goes
longitude) information from journalists and specialists.
missing. The KNN Classifier will then be trained using
cases that have already been registered. The crucial thing to keep in mind is that, in some cases, the
location of the found object is still unclear, even after
The prediction algorithm will be applied to every
becoming aware of reports. Given the location that was
single one of the aggregated photographs. If a match is
determined, the experiment only employed 29 examples in
discovered, it will be shown. There will be a display of
compliance with this work. The Gaussian filtering approach
every proven case. The ratio of the total number of errors to
is then used to remove noise and provide more exact just
the total number of test cases, with each instance of the data
face edges. Then, a logical operation is carried out on the
being used as a test case only once, is projected as the error
two previous output frames, and a noiseless face contour
rate obtained from all the tested cases.

2022 International Conference on Computer, Power and Communications (ICCPC) 563

Authorized licensed use limited to: Zhejiang University. Downloaded on October 24,2024 at 14:08:00 UTC from IEEE Xplore. Restrictions apply.
 frame is created for identifying edges that correspond to each frame. The scalar and vector distances between the
face video. four corner points of two subsequent frames must be
measured in order to track a human face from video. A
At the conclusion, there are four of these corner face's location and placement will change in the next frame
sites. Calculate top-top-right-bottom-left-top-right to create if one of the corner points moves.
a rectangle around the face and identify the facial shape in

IV. EXPERIMENTAL RESULTS

A. Registering New Cases

Registration of a new case is the initial stage. The GUI

application is created with PyQT5, which enables you to
gather all pertinent data and save it in the Postgres database.

Figure 4:Image Submission

C. Matching Cases
The next step is to match the case images and user
submitted images. To match KNN Algorithm is used.

Figure 3:Register New Case

B. Waiting for Users to submit images

We have just discussed "how new instances will be

recorded" up to this point; the next task is to match these
registered cases, but with whom do we do this? Where our
users come in is at this point. These users are regular folks
like you and I who wish to alter society. The regular people
will use a smartphone application to upload images of
persons they believe to be missing or discovered begging
while maintaining their anonymity. They value the
anonymity because they are concerned about potential
trouble from neighbourhood Gundas.

Figure 5:Matching Images

V. CONCLUSION

564 2022 International Conference on Computer, Power and Communications (ICCPC)

Authorized licensed use limited to: Zhejiang University. Downloaded on October 24,2024 at 14:08:00 UTC from IEEE Xplore. Restrictions apply.
One-shot learning has significantly improved the power of [12]. Criminals and missing children identification using face recognition
image identification. When properly applied, this and web scraping, S. Ayyappan and S. Matilda, IEEE ICSCAN 2020.
technology can be useful. It can also be applied to quickly
[13]. Find missing person using AI, International journal of Progressive
locate criminals in places like hotels and hospitals. The
Research in Science and Engineering, Vol. 2, No. 6, June 2021, Shefali
search for the missing people is well underway. Instead of Patil, Pratiksha Gaikar, Divya Kare, and Sanjay Pawar.
the time-consuming manual process of going through
databases for each image to see if it matches, our [14]. Effective Face Recognition System for Identifying Lost People,
technology quickly and accurately recognises faces. In the International Journal of Engineering and Standard Technology (IJEAT),
future, we plan to integrate this technology with ISSN:2249-8958, Volume-8, Issue-5 S, May 2019. Bharath Darshan Balar,
neighbourhood cameras to enable real-time face D S Kavya, Chandana M, Anush E, and Vishwanath R Hullipalled
recognition. The public cameras will provide frames to our
system continuously, and our system will continuously [15]. G. Kalyani, D. Meena, M. Lalitha, K. Swarna Bai Arnikar, K.RFID-
based missing person, Shirisha Seetasaikiran International Conference on
watch the frames.
Identity Systems, IEEE 2014Informatics, Electronics, and Vision
Conference (ICIEV)

[16]. P. A, N. G, V. K. R, P. P and S. R. R.V.T, "Stroke Prediction System

REFERENCES
Using Artificial Neural Network," 2021 6th International Conference on
Communication and Electronics Systems (ICCES), 2021, pp. 1898-1902,
doi: 10.1109/ICCES51350.2021.9489055.
[1]. B. Akinkunmi and P. C. Bassey. A Qualitative Approach for Spatial
Qualification Logic. International Journal of Artificial Intelligence & [17]. V. K. G. Kalaiselvi, A. Ponmalar, D. P. Rani, H. B and T. G.
Applications, 2017. Srivatsav, "FARM-A-CATION: A support for farmers," 2021 4th
International Conference on Computing and Communications
[2]. T. Bylander, J. R. Josephson, D. Allemang, M. C. Tanner, and The Technologies (ICCCT), 2021, pp. 97-101, doi:
Complex Computational Structure of the Abduction. Artificial intelligence 10.1109/ICCCT53315.2021.9711798.
in 2016

[3]. Cross-age face recognition using deep residual networks by Sarthak

Babbar, Navroz Dewan, Kartik Shangle, Sudhanshu Kulshreshtra, and
Sanjeev Patel. International Conference on Image Information Processing,
Fifth, IEEE 2019, (ICIIP).

[4]. Effective Face Recognition System for Identifying Lost People,

International Journal of Engineering and Standard Technology (IJEAT),
ISSN:2249-8958, Volume-8, Issue-5 S, May 2019. Bharath Darshan Balar,
D S Kavya, Chandana M, Anush E, and Vishwanath R Hullipalled

[5]. IRE Journals, Volume 1 Issue 12, ISSN: 2456- 8880, Birari Hetal,
Sanyashiv Rakesh, Porje Rohan, and Salwe Harish, "Android Based
Application-Missing Person Finder."

[6]. S. L. and do Ago Pereira N. by Barros. A reasonable framework to

examine variations on traditional planning is planning with abduction.
Pages 62 to 72 in Brazilian Symposium on Artificial Intelligence, 2018.

[7]. FaceNet: A Unified Embedding for Face Recognition and Clustering,

James Philbin, Dmitry Kalenichenko, and Florian Schrof, arXiv:
1503.03832V3 [cs.CV], 17 June 2015.

[8] . Face alignment by explicit shape regression[J], CAO X D, WEI Y C,

WEN F, et al. 2014;107(2):2887–2894; International Journal of Computer
Vision

[9]. Accurate object detection with joint classification regression random

forests [J], SCHULTER S, LEISTNER C, WOHLHART P, et al. 91(6):
591–599 in Fuel Processing Technology (2014).

[10]. Statistical Model-Based Face Pose Estimation[J]. GE Xinliang,

YANG Jie, LI Feng, et al. 2007;13(2):152-156 in Journal of Tianjin
University (English Edition).

[11]. CAO X, WEI Y, REN S, and others Utilizing local binary regressors
to align faces at 3000 frames per second [C]. Pattern Recognition and
Computer Vision (CVPR). IEEE, 2014: 1685-1692.

2022 International Conference on Computer, Power and Communications (ICCPC) 565

Authorized licensed use limited to: Zhejiang University. Downloaded on October 24,2024 at 14:08:00 UTC from IEEE Xplore. Restrictions apply.