Biometrics: Human Identification, AI Techniques, and Challenges Report

This report explores the application of biometrics for human identification, highlighting its importance in enhancing organizational security and efficiency. It delves into various AI techniques used in biometrics, including face recognition, fingerprint scanning, and eye lock systems, examining their functionalities, datasets, and performance metrics. The report analyzes the advantages of biometrics, such as improved employee verification and reduced fraudulent activities, while also addressing challenges like performance variations due to technical errors and behavioral changes. It focuses on CSL Limited as a case study and discusses the need for advanced technologies and AI integration to overcome these challenges. The report concludes that biometric machines are valuable for organizations and suggests systematic AI integration and continuous evaluation to improve the technology's effectiveness.

Biometrics based on human
identification

ABSTRACT
This present report is based in Biometrics based on human identification. Biometrics has
become one of the most useful of the organisation as it raises the productivity and efficiency of
the company. Biometrics has reduces many issues in the company such as unethical or fraudulent
behaviour of employees, ID swapping etc. Further, it has been identified that here are many types
of AI techniques which is used in biometrics. It includes machine such as face recognition,
fingerprint, eye lock, etc. that are installed in various devices and machines. These approaches
are very useful as it has raised the productivity and improve the payroll management to the
greatest extent but still there are various issues resulting in performance variations. Moreover,
the integration of biometrics with Artificial Intelligence (AI) has brought some improvement and
bit still the performance is varying. Lastly, it can be concluded that biometric machines are one
of the best advance technology which is used by organisation.

Table of Contents
ABSTRACT....................................................................................................................................2
INTRODUCTION...........................................................................................................................1
MAIN BODY..................................................................................................................................2
CHALLENGES & FUTURE RESEARCH.....................................................................................4
CONCLUSION................................................................................................................................4
REFERENCES................................................................................................................................6

INTRODUCTION
Background
Biometrics is becoming highly popular nowadays for the identification of human with the
rapid technological advancement. Biometric identifiers are the distinctive, measurable
characteristics used to label as well as describe individuals. Biometrics identifiers are often
categorized as physiological versus behavioural characteristics. It a technological and scientific
authentication method that is based on biology as well as used in information assurance (IA). It
also authenticates secure entry, information or access via human biological information like
fingerprints or DNA. It has brought huge improvements in the security of the organisation.
Artificial intelligence has played a great role as its integration has made very easy to identify and
verify the person. It has also improved the experience of employees working within organisation.
Significance
With the emergence of Biometric machines, organisations are enjoying many benefits.
Now, the employee verification, identification and management have become accurate, faster
and effective with this technology. Further, attendance tracking has also overcome employee
time theft along with prevent the dishonourable behaviour. It also helps in calculating the hours
of presence of employees in the company.
Problem Statement
Organisations face many issues if they do not have biometric machine within there
operations. Paper work increases, employees dishonourable behaviours is also arisen because of
attendance issues (Ruiz-Blondet, Jin and Laszlo, 2016). The entire payroll management is
negatively affected if biometric machines are not there. Security issues are frequent in the
absence of this technology. Further, unethical behaviour in organisation like ID Swapping,
undocumented access as well as regular badge checking.
Organisation Overview
The organisation chosen in this report is CSL Limited, this is an international speciality
biotechnology company that researches, develop, manufactures and market products to treat as
well as forbid serious human medical conditions. Company was founded in 1916 and
headquarters are located in Victoria.
MAIN BODY
1

In present times it has been important for organisation to maintain security of employees and
prevent any unauthorised access. This is because of rise in illegal activities and stealing of data
and info. Now, a new technology that is biometric is been used to identify and verify employee
identity. As said by Kumari and Vaish, 2015, with technological advancement biometric are
becoming popular in identification of human. It has allowed organisation to implement for
recognising humans. Biometrics is based on use of AI. The use of AI in different operational
process has made it easy to improve overall process. Moreover, AI has transformed overall
security systems. There have been a lot of improvements in security. Biometric is a machine or
computer science which is used as form of identification and access control. As stated by Gaber
and et.al., 2016, there are many types of AI techniques which is used in biometrics. It includes
machine such as face recognition, fingerprint, eye lock, etc. that are installed in various devices
and machines. Each of the AI technique differs from one another. Furthermore, the features
installed are unique which enables in securing that device. In addition integration of biometric
with AI has ease process to verify and identify person identity. The use of AI in biometric has
highly improved its performance. Alongside, various types of dataset are used which enable in
identification process. The performance is plotted in ROC or DET form. Apart from it, there is
need to get things done in proper ways. With this, there is decline in system failure and rejection
rate. The approaches are described below :-
Face recognition – It is the most secure and common technique of biometric. Here, a person
face is identified and verified. The sensors detect expression and different angles of face of a
person. Also, face shape, cells, etc. are scanned that include ear, nose, head, etc. the technique
create a visual shape of overall face. Then, it is matched with algorithm of specific person. The
algo are different and identify face. This technique works in real time by gathering and
identifying person. It uses UMDF dataset where a large set of data is stored of individual face.
So, if a person face ID is recognised from its face expression, colour, etc. there are over
3,700,000 frames in dataset including annotated faces. Moreover, there are many algorithms in
dataset through which identification is done. The performance of face recognition is done with
help of algorithms (Johnston and Weiss, 2015). Here, correlation matching, principal
component analysis and local feature analysis is used. It enables in ensuring that person face is
verified. Sometimes, performance of face recognition is impacted due to technical errors. This
can result in getting variation in results. Thus, the verification process is affected and person
2

identified may be wrong one. Its process is quick as facial expressions are scanned easily through
frames. The algorithms are applied one by one and matching is done. In this way there are few
errors in face recognition. Another major issue in performance of face recognition is change in
behaviour. A person facial expression is affected by change in behaviour. Due to this, face
recognition technique does not work. Hence, its performance s highly affected.
It is both 2D and 3D depending on biometric used. In article multi modal is compared
with default multi sample. So, multi modal 2D +3D face matched rank recognition rate is 95%.
Multi sample face with tow sample rank recognition rate is 92.9%. However, multi sample 2D
face with 4 sample rank recognition rate is 96%. Also, multi modal orthogonal biometric is used
in which 2D and 3D ear image is recognised. Moreover, there are various types of multi modals
used such as independent, collaborative, etc. Thus, the results are as follows :-
Fingerprint – This is also a technique of biometric in which individual identify is verified and
identified based on fingerprints. It is mostly used technique in organisation and for person use as
well. The technique is fully computerised and automated. The dataset used in this is multimodal
biometric data BIOMDATA. It is connected with software which identify fingerprint. However,
its performance is effective as each person fingerprint is scanned. The errors are occurred due to
optical and capacitive sensor. Also, once dataset is filled with both fingerprints the verification
match between all combinations is stopped. Hence, due to equal errors rate there is degradation
in performance of fingerprint recognition. Moreover, sometimes in spite of frequent errors in
fingerprint the software fails (Semwal, Raj and Nandi, 2015). Also, it takes time in verifying as
3

fingerprint may not be recognised easily in one attempt. In addition, if there is change in finger
shape the biometric is not able to scan and identify person.
Eye lock – this is also an advance technique of biometric where a person is identified on basis of
their eye. It is a process in which retina is scanned that provides output of person identity.
Furthermore, eyesight, eye colour, etc. is completely scanned and then overall verification
process is completed. The eyes are most reliable body part in human. This is because every
person eye sight and retina is different. The algorithms designed are complex as it ensured that
technique used is very effective. The dataset used in it comprises of advance machine learning
and algorithms. The performance of eye recognition varies to a great extent. It contains multiple
characteristics which makes it difficult to verify. Due to this performance is impacted. In eye
lock biometric performance depends on type of dataset used. It has been observed that
sometimes eye lock a person eye is not verified due to change in condition of eye. It takes time to
identify identity. In this case performance is affected.
According to Abo-Zahhad, Ahmed and Abbas, 2016, these all biometric approaches are
used in identification process. In recent times it has been integrated with AI. Hence, with use of
AI process has become easy. Now, separate dataset is used with technique. Moreover, the
performance of techniques is enhanced with use of algorithms. However, variations in
performance are identified which makes it easy to eliminate it. There is need to evaluate
performance of each technique. Henceforth, pre defined data set integrate with authentication.
CHALLENGES & FUTURE RESEARCH
There are various challenges that organisation face with the implementation of Biometric
human identification machines. CSL company is also facing some issues in using different
approaches of Biometric machines. Here are the issues-
Fair Recognition Challenge: this approach is recognised as the secure technique of
biometric but the change in the behaviour of employees result in change in behaviour of
individual which sometimes does not able to recognise. This error affects the performance of
face recognition approach of biometric (Sprager and Juric, 2015).
Fingerprint Challenge: this technique is also used in organisation for attendance of
employees which sometimes is not work well which influenced the performance. Change in the
4

shape of finger or any injury on the finger do not bring right result as system unable to scan and
identify the employee.
Eye Lock Challenge: this approach of biometric is one of the advance technique that
affect the performance at organisation. Eye Lock is very time consuming process that make
delay in other employees to enter in the premises. Eye lock challenge result in having long queue
on the gate which also result in late attendance of other employees.
Thus, this are the above challenges that organisation faced with the implementation or
application of biometric approaches that needs to be resolve so that performance variation do not
occur. So, in order to reduce the variations, CSL should use more advance technologies and keep
them updated with the new software (Prakash, Kumar and Mittal, 2018). Further, proper
integration with Artificial Intelligence (AI) should be there so that these issues or challenges can
be overcome to the greatest extend. In addition to this, close monitoring and evaluation should
also be there in the company regarding biometric machines which also overcome these issues.
Lastly, it can be said that, the performance variation can be improved by systematic and effective
integration of AI technology with Biometric machines. There are also many variations coming in
identification process. This is because of change in algorithms and copying of data over it. Due
to this it algo in dataset are not matching with biometrics. However, variations are also due to
technical errors or faults in techniques.
CONCLUSION
From the above report, it has been identified that biometric machines are one of the best
advance technology which is used by organisation. It has provided dynamic solutions for
business owners as there are many issues which they faced in the absence of biometric machines
such as fraudulent behaviour of employees, ID swapping, burden of paper work, tough payroll
management. Biometric is a machine or computer science which is used as form of identification
and access control. Further, finger print, Eye Lock, face recognition etc. are the most common
technique applied in the organisation to increase the productivity and efficiency of the company
performance. Further, it has been identified from the report that, thee approaches creates
variations in the performance because of some issues like it is time consuming, change in
behaviours of individual and injury or change in shape of fingers affect the indemnification of
person.
5

6

REFERENCES
Books and Journals
Abo-Zahhad, M., Ahmed, S.M. and Abbas, S.N., 2016. A new multi-level approach to EEG
based human authentication using eye blinking. Pattern Recognition Letters. 82. pp.216-
225.
Gaber, T. And et.al., 2016. Biometric cattle identification approach based on weber’s local
descriptor and adaboost classifier. Computers and Electronics in Agriculture. 122. pp.55-
66.
Johnston, A.H. and Weiss, G.M., 2015, September. Smartwatch-based biometric gait
recognition. In 2015 IEEE 7th International Conference on Biometrics Theory,
Applications and Systems (BTAS) (pp. 1-6). IEEE.
Kumari, P. and Vaish, A., 2015. Brainwave based user identification system: A pilot study in
robotics environment. Robotics and Autonomous System. 65. pp.15-23.
Prakash, C., Kumar, R. and Mittal, N., 2018. Recent developments in human gait research:
parameters, approaches, applications, machine learning techniques, datasets and
challenges. Artificial Intelligence Review. 49(1). pp.1-40.
Ribaric, S., Ariyaeeinia, A. and Pavesic, N., 2016. De-identification for privacy protection in
multimedia content: A survey. Signal Processing: Image Communication. 47. pp.131-151.
Ruiz-Blondet, M.V., Jin, Z. and Laszlo, S., 2016. CEREBRE: A novel method for very high
accuracy event-related potential biometric identification. IEEE Transactions on
Information Forensics and Security. 11(7). pp.1618-1629.
Semwal, V.B., Raj, M. and Nandi, G.C., 2015. Biometric gait identification based on a
multilayer perceptron. Robotics and Autonomous Systems. 65. pp.65-75.
Sprager, S. and Juric, M., 2015. Inertial sensor-based gait recognition: A review. Sensors. 15(9).
pp.22089-22127.
7