TELONE CENTRE FOR LEARNING

DIPLOMA IN SOFTWARE ENGINEERING

FACIAL RECOGNTION ATTENDANCE SYSTEMI: A CASE

OF KUNTEM TRADERS MICROFINANCE HARARE

BY JOHN J CHIKODZE (T2056371Z)

ID NUMBER: 63-2198236 V 47

CLASS: SOFTWARE 2020

MAIN SUPERVISOR: MISS CHIBHABHA

CO-SUPERVISOR: MR SUWALI

1|Page
1.0 Introduction.....................................................................................................................................3
1.1 Background......................................................................................................................................4
1.2.0 Problem Statement......................................................................................................................4
1.2.1 Proposed Solution........................................................................................................................4
1.3.0 Aim...............................................................................................................................................5
1.3.1 Objectives.....................................................................................................................................5
1.4Hypothesis........................................................................................................................................5
1.5 Significance of the project...............................................................................................................5
1.6.0 Limitation of the study.................................................................................................................6
1.6.1 Delimitation of the study..............................................................................................................6
1.7.0 Methodology................................................................................................................................6
1.7.1 Flow chart (Activity diagram)........................................................................................................7
1.7.2 Design tools..................................................................................................................................8
1.8.0Feasibility study.............................................................................................................................8
1.8.1 Technical feasibility......................................................................................................................9
1.8.2 Hardware requirements...............................................................................................................9
1.8.3 Software requirements...............................................................................................................10
1.8.3Economic feasibility.....................................................................................................................10
1.8.4Operational feasibility.................................................................................................................10
1.9 Gantt chart....................................................................................................................................11
CHAPTER 2.......................................................................................................................................12
LITERATURE REVIEW........................................................................................................................12
2.0 Introduction...............................................................................................................................12
2.1.0 Machine Learning...................................................................................................................15
2.1.1 Supervised Learning................................................................................................................15
2.1.2 Unsupervised Learning...........................................................................................................15
2.1.3 Tobacco Market......................................................................................................................16
2.2Conclusion..................................................................................................................................17

2|Page
 CHAPTER 1
1. INTRODUCTION
Face Recognition is a popular image processing technology because of its widespread usage.
Face recognition may be used to identify students in an institution for attendance purposes.
The maintenance and evaluation of attendance records is critical in every institution`s
performance review. The aim of creating an attendance monitoring system is to automate the
conventional method of taking attendance. With less human interaction, the Automated
Attendance Management System conducts the everyday tasks of attendance marking and
review. When the intensity is greater, the traditional form of attendance marking becomes
very time consuming and complicated. Automation of Attendance System has an advantage
over conventional methods in that it saves time and can also be used for monitoring. This also
aids in the prevention of false participation. Other biometric techniques, such as those
mentioned below, can also be used to computerise the attendance process: 1. Log Book entry

2. Fingerprint based System

3. IRIS Recognition

4. RFID based System

5. Face Recognition

Facial recognition is the most unique, efficient, precise, and cost-effective of all the
techniques described above

3|Page
 2. BACKGROUND
Tobacco

3. PROBLEM STATEMENT
 It becomes more difficult to mark attendance for each student when there are so many
students in an organisation and it is a time consuming one. The Existing system at the
institute is manual entry for the students. This system faces the issue of wasting time
and it becomes complicated when the strength is more. It is very tedious job to carry
out the attendance in log books and to maintain the records.

4. PROPOSED SOLUTION
 The approach suggested in this project is to use facial recognition technology
to monitor attendance. The computer captures camera, video streams and
senses faces in image format. The identified faces will be linked to the student
database, and the attendance will be recorded in an Excel spreadsheet. Using
these Excel boards, we will create a graph that displays the average attendance
of the whole class/individual student.

1.3.0 Aim
To create an Auto-mated Attendance System, Touchless Sign-in System (Post pandemic
requirement), Viability with cloud and which reduces manual effort.

1.3.1 Objectives
 To create a Facial Recognition Attendance System that can be used to capture
attendances of students in an institution.
 To develop an interface that can be used by Lectures to monitor the students
attendances patterns.
 The system is aimed to be efficient, time-saving, simple, and easy to use.

1.4Hypothesis
H0 Null hypothesis: The Facial Recognition Attendance System will not be able to capture
faces and recognise captured faces.

H1 Alternative hypothesis: The Facial Recognition Attendance System will be able to capture
faces and recognise captured faces.

4|Page
1.5 Significance of the project
This system is of great significance to the students because it enables students to reduce the
risk of losing the Pass cards and Long ques that is quick check-in.

Accuracy: Facial recognition technology is highly accurate in identifying individuals,

which makes it an efficient and reliable method for attendance tracking.

Time-Saving: The system eliminates the need for manual attendance tracking, which can
be time-consuming and prone to errors. Facial recognition attendance system can save time
and resources for the institution.

Security: The System can improve security by accurately identifying individuals and
preventing unauthorised access to a location or event

Convenience: The system is convenient for both administrators and attendees. Attendees
can quickly check-in and administrators can easily access attendance records.

1.6.0 Limitation of the study

The researcher will face a challenge on accessing prices due to some information restrictions
laid down according to the company policies ,as well as financial constraints on cost incurred
to gather all the needed resources for the project to be successful. Tobacco is also affected by
other factors that can be put in place in future updates.

1.6.1 Delimitation of the study

The researcher is mainly focusing on developing a Tobacco price prediction based on pre-set
data set for Tobacco marketing board, Boka auction floor and Premier auction floor, whether
the system is effective or not effective when farmers are making decisions on tobacco
farming and thereby at the same time ignoring some Tobacco floors that could have been
involved in the research.

1.7.0 Methodology
In this study l will mainly focus on the quantitative research and the data set for rainfall and
sunshine is going to be obtained from the Meteorological services department of Zimbabwe. I
will also use the overall demand, prices previously set by the government and train the

5|Page
system using that data set. The system will train and learn from previous prices set by the
government using linear regression algorithm and predict future price. I will also visit the
operational environment (Government Tobacco marketing board) personally to acquire the
desired information. The system will also be open and generic and can be used by any other
auction floor and is going to have pre-set data set for Tobacco marketing board, Boka auction
floor and Premier Tobacco floor.

1.7.1 Flow chart (Activity diagram)

face

Fig 1. 1 Activity diagram.

6|Page
1.7.2 Design tools
 Intellij IDE
 phpMyadmin (mysql database)
 Excel spreadsheet
 Bootstrape framework
 Linear regression algorithm

Languages
 C++
 JavaScript

1.8.0Feasibility study

Technical feasibility

Economic feasibility

Feasibility study Operational feasibility

Fig 1. 2 Feasibility study diagram

7|Page
As the name implies, a feasibility study is an analysis of the viability of an idea. The
feasibility study focuses on helping answer the essential question of “should we proceed with
the proposed project idea?”. It describes a preliminary study undertaken to determine and
document a project’s viability. The results of this analysis are used in making the decision
whether to proceed with the project or not. It shows whether the project create a technical and
economically feasible concept. This is the process that will identify any possible problems
that might occur between the acceptance of the product with the consumer and how profitable
the system is. Feasibility study also establishes whether the pharmaceutical department has
the capacity to support the project in terms of technical, economical and operational
conditions.

1.8.1 Technical feasibility

Technical feasibility aims at analysing the Web-based system from a technical perspective,
that is whether the required technology is available or not and also the availability of the
required resources. Technical feasibility of the proposed system may be done by just looking
at the software and hardware available and analysing them without necessarily building the
system. The technology for the proposed system is shown below:

1.8.2 Hardware requirements

This is the analysis of the hardware and infrastructure that is needed for the implantation of
the project. Minimum specification of computer to be used are 3 GHz processor speed, 8GB
RAM, 500Mb Hard Disk Drive, Windows Operating System (64-bit windows 10) since the
system is web based.

Table1. 1 Hardware Required

Description Quantity Availability

Laptop/desktop 1 yes

The table above shows the availability of the hardware requirements of the proposed
system, therefore making the proposed system technically feasible.

8|Page
1.8.3 Software requirements
Table1. 2 Software Required
Description Quantity Availability
Windows 10 1 yes
Operating
System

MySQL 1 yes
database

Pycharm 1 yes

Google Chrome/ 1 yes

Flask framework 1 yes

The above table shows the availability of the software requirements of the proposed system,
therefore supporting the idea that the system is technically feasible.

1.8.3Economic feasibility
Economic feasibility is the worthiness of a project or the return on investment computation.
The economic feasibility of this project is measured by the extent to which the utility value
received during the entire project or product life, exceeds the proper costs incurred in creating
the utility accounting revenue, while costs (also over the project life) are equivalent to the
accounting expense. The proposed system is economically feasible since resources to
facilitate the project’s success are available and can be acquired for free. All the hardware
equipment, software and labour required are available.

1.8.4Operational feasibility
Operational feasibility is a measure of how well a proposed system solves the problems, and
takes advantage of the opportunities identified during scope definition and how it satisfies the
requirements identified in the requirements analysis phase of system development. It answers
question “will the system work?” and how well the system would fit into the current business
structures and once implemented how useable would it be. Operational feasibility is
dependent on human resources available for the project and involves projecting whether the
system will be used if it is developed and implemented. After analysing these operational
requirements, we see that the proposed system is operationally feasible.

9|Page
1.9 Gantt chart

Project work plan (Gantt chart)

ACTIVITY Week Week Week Week Week Week Week

1–2 2– 4 4–5 6–8 9– 11 12– 13 14-15

Problem identification

Literature review

Research

Design and Coding

Testing

Evaluation

Deployment

Conclusion

10 | P a g e
 CHAPTER 2
LITERATURE REVIEW
2.0 Introduction
This chapter will mainly discuss the study of how price prediction systems have been carried
out by, other authors in the similar area of the present study. This chapter will focus on the
discussion of theoretical and practical (empirical) views of previous studies done in
prediction systems. The study also covers a wide range of issues related to tobacco in
different subject areas such as demand, supply, trade of tobacco leaves, demand for
cigarettes, cigarette advertising, cigarettes taxes, social costs of smoking, economic
significance of tobacco industry, and farm issues associated with tobacco control. Most of the
tobacco produced in the world is not consumed as a final consumer product, but is used to
produce tobacco products.

In 2012, [1] Reza Ghodsi, Ruzbeh Mirabdollah Yani focused on predicting wheat production
Using Naïve Bayes Algorithm. Wheat forecasting was done using Naïve Bayes Algorithm to
predict future price of the crop. The Argument was that they had few data that is why Naïve
Bayes was used. Naïve Bayes classifier assumes that the presence/absence of a particular
feature of a class is unrelated to the presence/absence of another feature, given the class
variable. For Example, crop is considered to be wheat if it is brown and small grain. Even if
these features depend on each other or upon the existence of the other features, a naïve Bayes
considers all of these properties to independently contribute to the probability that the crop of
high quality. Advantage it was fast to be implemented since Naïve Bayes Algorithm is one of
the simple to implement as they say. Another advantage was that a Naïve Bayes classifier
converge quicker than other models such as logistics regression, which suits when they say
they had less data to train. The used method was successful in that it was almost 90%
accurate in the outcomes it gave. The main additional and advantage feature of this was that it
was able to predict future trends of the pricing for next five years.

11 | P a g e
In 2012, [5] Raquel Salazar Moreno and Oscar Zeerecero Salazar made a maize prize
behaviour system using multilayer perceptron. When this was performed, a multilinear
perceptron was used to analyse the behaviour of maize per season of maize production. To
build this, a neural network model was made to simulate the price behaviour of corn based on
prices and related products and the international price of corn. Accuracy of the multilayer
perceptron performed on the prize behaviour analysis was 46.06%. Aim was to analyse the
price behaviour in terms of what could make the price goes up or down, it was achieved. The
multilayer perceptron also was able to prove that maize is directly affected by amount of
rainfall per season. This study was also able to generate a policy of joint support products in
determining which price has the greatest effect on the corn price. Due to use of perceptron
lack of error correction, they suggested that for future use, they would apply the multilayer
perceptron with back propagation algorithm for correction of errors (weights).

In 2007, [3] Veenadhari Suraparanju connected Choice Tree Calculations for Soybean

Productivity System. A choice tree could be a flow-chart-like tree structure, where
each inside hub indicates a test on a property, each department speaks to classes
or course disseminations. [2] The beat most in the tree is the root hub. In arrange to classify
an obscure test are tried against the choices tree. A way is followed from the root to a
leaf hub that holds the course expectation for that sample. The ponder illustrated the
potential utilize of this handle for extricating valuable data. The rules Shaped from
the choice were supportive in anticipating the conditions mindful for high
or more efficiency of the soybean. The most challenge that was confronted from this method
was the algorithm’s complexity. The strategy was not effectively actualized due to that
decision tree was overfitting (the arrange had retained the preparing illustrations but it was
not able to generalize to modern situations).

12 | P a g e
Cotton Price Forecasting in Major Producing States of India by Ashwini Darekar1 and A
Amarender Reddy (2017). In this paper, the mentioned authors developed a cotton price
forecasting system. The system was developed due to cotton high price fluctuation rate,
which depends on the global business cycles. It was investigated that cotton farmers
encounter high price risk, hence the importance of cotton price forecast for the benefit of
farmers as well as millers who purchase cotton. The time series data on monthly price of
cotton required for the study was collected from the AGMARKNET website from January,
2006 to December, 2016 to forecast prices for kharif 2017-18 year harvest months. ARIMA
model was used to predict the future prices of cotton. Model parameters were estimated using
the R programming software. The performance of fitted model was examined by computing
various measures of goodness of fit viz., AIC, SBC and MAPE.

In 2017 Nathan Mitchell merged production and weather data and used time series analysis
and machine learning techniques to predict global coffee prices. It was noted with concern
that cotton prices fluctuates and Coffee producing countries collaborated with UN to trey to
overcome the problem in 1962. An International Coffee Agreement that set up an export
quota system was created to help stabilize prices, and the International Coffee Organization
to help regulate the system. The target was to promote the spread of industry knowledge and

improve economic conditions for farmers. Standard data cleaning and formatting was
perfomed converting strings into date time values, filtering the weather stations based on their

latitude and elevation. Random forests model was used by building decision trees and
randomly removing features to minimize overfitting. Large number of trees and cross-
validation were also used, which are also ways to minimize the chances of overfitting, and
minimizes errors when new data is shown to the model.

13 | P a g e
Lubna A. Gabralla1, Hela Mahersia, and Ajith Abraham investigated an Ensemble
Neurocomputing Based Oil Price Prediction. Daily data from 1999 to 2012 were used to
predict the West Taxes, Intermediate. Data were separated into four phases of training and
testing using different percentages and obtained seven sub-datasets after implementing
different attribute selection algorithms. Three types of neural networks were used: Feed
forward, Recurrent and Radial Basis Function networks. Eventually a good ensemble neural
network model was formulated by the weighted average method. Empirical results illustrated
that the ensemble neural network outperformed other model.

2.1.0 Machine Learning

Many Machine Learning Techniques have been explored for Crop price predictions, ANN
and Linear Regression are two widely used machine learning algorithms for predicting
tobacco price predictions and tobacco market index values. A literature survey of supervised
and un-supervised machine learning methods applied in tobacco market analysis is presented
below

2.1.1 Supervised Learning

Supervised learning is the machine learning errand of learning a function that maps an input
to a yield based on case input-output sets. It deduces a work from labelled preparing
information comprising of a set of preparing cases. In directed learning, each case may be a
match comprising of an input question and a wanted yield value.

2.1.2 Unsupervised Learning

Unsupervised learning is a sort of machine learning algorithm utilized to draw inferences
from datasets comprising of input information without labelled responses. The most common
un-supervised learning strategy is cluster analysis, which is utilized for exploratory
information examination to discover covered up designs or grouping in information.

14 | P a g e
2.1.3 Tobacco Market
Boka Tobacco Floors (BTF) has got both the systems and space to facilitate the buying and
selling of tobacco. Farmers bring their bales to BTF who receive, weigh and lay on the floor
the tobacco. The buyers then bid for each bale offering a specific price. The highest bidder
wins. The auctioneers and related staff are provided by BTF.

A sale is concluded when the Farmer accepts the offered price after which BTF dispatches
the bales to the merchants. Sometimes merchants do not want to participate in an open bid-
ding process. Under such circumstances the merchants would provide inputs to specific
farmers. BTF provides space and systems for these merchants as well.

Given tobacco’s importance, the Government established a regulatory board, the Tobacco
Industry and Marketing Board (TIMB) whose main mandate is to ensure sustainable
production and orderly marketing of tobacco.

The impact of rainfall variability on rain-fed tobacco in Manicaland province of Zimbabwe

(2011) by Steven Jerie and T. Ndabaningi (Meteorological Services Department, Harare).
The study focused on the impact of rainfall changes on tobacco production in Nyazura district
of Manicaland. Tobacco is one of the cash crops that is greatly affected by climate changes.
Questionnaires and interviews were used for data collection from farmers and agricultural
research extension officers. The conclusion was that rainfall variability has a great influence
to the yield of tobacco.

Assessing the impact of climate change on the suitability of rainfed flu cured tobacco
(Nicotiana tobacum) production in Zimbabwe (2013) by Chemura Abel1, Kutywayo
Dumisani, Mahlatini Precious, Nyatondo Upenyu & Rwasoka D. The mentioned authors
outlined that Tobacco production in Zimbabwe is profitable due to supportive ecological and
socioeconomic conditions. The challenge was that ecological conditions vary with changes to
climate conditions. The crop suitability modelling was carried out using Maxent species
distribution model. They also used current and projected climatic variables to understand the
changes in potential distribution of rainfed tobacco production in Zimbabwe.

15 | P a g e
2.2Conclusion
The aim of implementing this system in place is to save time and money by reducing
manpower. This method demonstrates a deeper understanding of the algorithm and a rigorous
approach to accurately recognising users. The end result shows the system's ability to deal
with improvements in face posing and projecting, as well as changes in room. According to
facial recognition through machine learning, face detection addresses the problem of change
in environment when the original image is transformed into a HOG representation that
captures the key features of the image independent of image brightness. Local facial
landmarks are taken into account for further processing in the face recognition system. The
caught face is then encoded, yielding 128 measurements, and the best face identification is
accomplished by removing the person's name from the encoding. The outcome is then used to
build an excel spreadsheet. Currently the System has attained an accuracy up to 91%.

16 | P a g e