AN AI BASED AUTOMATED SUPERMAKET SYSTEM
Abstract – In this contemporary society, many sections are being modernized due to a trend towards technology. Nowadays, it is impossible to find a person without a smartphone. People use smartphones to get their day-to-day activities done easily. Automated supermarkets can be pointed out as one such field where smartphone is used for. Instead of traditional outdoor
shopping from one place to other, the art of shopping has been developed in various ways using technology. We started to find out a strategy to facilitate the customer to purchase without any rush. The aim of this is to facilitate the customer with best technology from the moment he enters the supermarket and to the end of his transaction without any rush.
Here we expect to develop an online system focused on AI (Artificial Intelligence). The proposed automated supermarket system performs using an analyzing technique in machine
learning algorithms. When a customer enters a supermarket, he/she is identified through this Mobile Application (App), and after supplying him/her a cart he/she goes to the relevant shelf.
Using the FP-Growth algorithm which comparison of earlier purchases, new suggestions of commodities are sent to the App. The mobile application analyses the amount of nutrition
contained by the previous shopping time and compares it with the amount of nutrition contained by the current shopping time using Singular Value Decomposition (SVD) and Content-Based Filtering with the introduction of a Content Suggesting Algorithm. Then the relevant bill is displayed in the App. The mobile application uses the sorting method, CAS (Compare and swap) algorithm, classification for suggesting the suitable bank promotion to the customers by analyzing products, purchasing behavior, and bank cards. Meanwhile, the staff of the supermarket is informed about the details of commodities purchased by a customer via the Desktop App with the help of Compare and Swap (CAS) algorithm. As a result, by this expected strategy, not only the customer but the employer are
also facilitated. Although device-based automated supermarkets are established in various places, they seem to be unsuccessful. The cause for this would be the cost for the devices and
maintenance. Therefore with the implementation of our system, customers become convenient for shopping and it assists uplifting the supermarket profits well.
Keywords—Artificial Intelligence, Smart Supermarket, Mobil
Authors —D.N.U Pullaperumage,N.G.S.Y.K Gamage, H.C.P.P Hewavitharana, W.M.P.S Wicramasingha , Suranjini Silva
AUTOMATIC INTONATION RECOGNITION OF SINHALA LANGUAGE TO DETECT SPEECH
IMPAIRED IN YOUNG CHILDREN
Abstract – Speech recognition is a better approach to extract,
identify and analyze specific features in the voice. Further, this
can be utilized to analyze tone, accent and word behaviors such
as whether the speaker has made a statement, question or
command. This approach comes with a system to recognize
Sinhala speech of children between 1 to 6 years based on
intonation using Machine Learning. The model has been trained
using Neural Network by extracting MFCC features, chroma
and Mel based on fundamental frequency(f0). The final system
can record, analyze and generate a report on the word behavior
of the child which can be used by the speech pathologists to track
the improvement of the child. Due to the unavailability of a
proper database on Sinhala utterances of children, data has
been collected manually in a real environment. Hence, data has
been filtered and labelled accordingly. Finally, in the training
phase, the system has 85% accuracy and overall accuracy of
90% in the testing phase. With the gap of Sinhala related
intonation recognition of children, this approach can be further
utilized in other research areas as IoT, smart devices and,
speech and voice recognition by other researchers.
Keywords—Machine Learning; Speech recognition; Sinhala
Intonation; MLPClassifier
Authors —Chathurindu Wickramaarachchi, Veerandi Kulasekara, Koliya Pulasinghe, Vijani Piyawardana
DESIGN AND SELECTION OF A SOLAR POWER DRIVEN WATER PUMPING SYSTEM FOR
PROPOSED WATER POCKETS IN POLGAHAWELA, POTHUHERA, ALAWWA INTEGRATED WATER
SUPPLY PROJECT
Abstract – The main objective of this design is to introduce
a sustainable power solution for remotely located pump houses
of the National Water Supply & Drainage Board. The same
pump houses are in the locations where continuous power
interruptions happen in the CEB (Ceylon Electricity Board)
power supply. Further, some of the newly planned pump houses
are in the locations where the difficulties are available in getting
the CEB power supply. Hence, this design is carried out to
provide uninterrupted power supply to the mentioned pump
houses and at the same time, to provide a sustainable solution
by reducing incurring CEB electricity costs or Fuel costs for
Diesel Generators. This in turn gives the benefit of Return-OnInvestment (ROI) the NWSDB while giving an uninterrupted
service to the consumers. One of the 8 numbers of Water Pocket
solution pump houses will be selected to carry out the design and
the selection of the relevant equipment will be commenced based
on the results of the design calculations.
Keywords—
Authors —D.M.S.A Dissanayake
HUMANOID CARE ROBOT FOR NURSING HOME IN SRI LANKA
Abstract – Humanoid Care robots are used in hospitals and
houses to care for and assist inclined human beings, which
includes the aged, children and those with physical and
intellectual disabilities. The paper has two predominant
objectives such as have a look at the challenges springing up from
the use of elderly care humanoid robots at nursing homes as well
as to discuss how the layout of elderly care humanoid robots can
deal with these challenges. To recognize higher the real-global
capability of robot-based assistance, we undertook a case study in
a nursing home involving agencies of elder citizens, caregivers
and managers as stakeholders. We identified each, enablers and
barriers to the ability implementation of robotic structures.
Development approaches are data information collected to
propose the model, the proper way to interconnect electronic
components, practical limitation, system tasks, functional system
research, sample system design. This article evaluates the
introduction of the new humanoid system and its background,
explains the literature review, theoretical approach, methods
resources used, activities carried out for the research
methodology. achievements and methods adopted to meet these
objectives, the design and implementation and final design of the
system.
Keywords—Autonomous Mobile Robots, Nursing Home, Elder
care, Humanoid Robot
Authors —Mr.Vadivelthasan Jayathas, Ms. A. H. S. Shamini Dinuka
CHILD’S AGE RANGE PREDICTION USING SINHALA SPEECH RECOGNITION SYSTEM
Abstract – — This research aims at determining the age range of a
child who speaks to the computer by analyzing voice
characteristics and acoustics features. To screen speech
impairment of children between 6 months to 72 months, it is
important to implement a system that can recognize their native
language to predict the age range. The implemented system
generates a highly accurate report for speech pathologists to use
as their second opinion when diagnosing a child. In this
research, Multilayer Perceptron based neural network is
proposed to identify the age group of a child who speaks Sinhala
as their native language. The age recognition system’s
performance is determined by the speech features that have
been used by the child. Mel Frequency Cepstral Coefficients
(MFCC) are chosen to capture the difference because it is a
useful function for speech recognition. A blend of pitch and Mel
Frequency Cepstral Coefficients (MFCC) features were used to
increase recognition rates even further. The final method
implemented shows a 77% accuracy rate of overall
identification of age range for a child.
Keywords—Speech Recognition System (SRS), Mel
Frequency Cepstral Coefficients (MFCC), Speech pathologist,
Age range, Speech impairment
Authors —Ashimi Kathriarachchi, Veerandi Kulasekara, Vijani Piyawardana, Koliya Pulasinghe
CHILD’S AGE RANGE PREDICTION USING SINHALA SPEECH RECOGNITION SYSTEM
Abstract– This research aims at determining the age range of a
child who speaks to the computer by analyzing voice
characteristics and acoustics features. To screen speech
impairment of children between 6 months to 72 months, it is
important to implement a system that can recognize their native
language to predict the age range. The implemented system
generates a highly accurate report for speech pathologists to use
as their second opinion when diagnosing a child. In this
research, Multilayer Perceptron based neural network is
proposed to identify the age group of a child who speaks Sinhala
as their native language. The age recognition system’s
performance is determined by the speech features that have
been used by the child. Mel Frequency Cepstral Coefficients
(MFCC) are chosen to capture the difference because it is a
useful function for speech recognition. A blend of pitch and Mel
Frequency Cepstral Coefficients (MFCC) features were used to
increase recognition rates even further. The final method
implemented shows a 77% accuracy rate of overall
identification of age range for a child.
Keywords—Speech Recognition System (SRS), Mel
Frequency Cepstral Coefficients (MFCC), Speech pathologist,
Age range, Speech impairment
Authors —A.M.N.C. Attanayake, W.G.R.U. Hansamali, R. Hirshan, M.A.L.A. Haleem, M.N.A. Hinas
MOBILE APPLICATION TO IMPROVE MENTAL HEALTH OF PERSONS WITH DEMENTIA
Abstract -Dementia is a disorder because of illness of the
cerebrum, as a rule of a persistent or reformist nature, where
there is the unsettling influence of different higher cortical
capacities, including memory, thinking, direction, cognizance,
figuring, learning ability, language, and judgment. To address
these perplexities, this research has designed a mobile
application to help caretakers, family members and friends to
have better interaction with the people with Dementia and take
care of them to improve the quality of living for the cognitively
disabled. The specialty of this application is brain recovery
activity to help the person with Dementia to recover their
memory up to some extent. Also, system provides an inbuild
music player with pre-loaded music list which is suitable for the
persons with Dementia and application model trained as a
chatbot understand the user’s questions and chatbot responds.
The paper further looks at the event’s outcome, as well as the
benefits, feedback, and potential work for the application and
its users. The article emphasizes the artefact’s effectiveness and
how it accomplishes its goals and priorities. The researcher
discovered that the chatbot has a better success rate of accuracy,
with in the range of 80-90% of accuracy showed 60% success
rate out of general responses. This importance demonstrates
that, to achieve a higher accuracy score, the chatbot must also
be accurate. The accuracy of the chatbot and enhancing the
available features in the mobile healthcare application, also the
researcher planning to add new features to the system, as a
future work.
Keywords—Dementia, Counselling, Application,
Alzheimer’s, Relaxation Methods, Music, Artificial Intelligence,
Machine Learning
Authors —Nimesh Mendis, Vibhavi Attigala
INVESTOR ORIENTED STOCK MARKET PORTFOLIO ANALYZER AND MANAGEMENT PLATFORM
SUPPORTED BY STOCK PRICES PREDICTION FOR COLOMBO STOCK EXCHANGE OF SRI LANKA
Abstract -Over the past few years various studies have been
conducted to develop an optimum stock market related portfolio
management platform that will assists investors to actively
perform the portfolio management process. Risk and level of
investor participation is considered to be one of the challenging
aspects identified for optimum portfolio management. Along
with portfolio management, stock price prediction is one of the
key contributing factors that helps an investor to arrive midand long-term strategic investment decisions. Various deep
learning concepts are evaluated to determine the most accurate
algorithm to implement the stock price-based prediction system.
Currently Colombo Stock Exchange have identified a desperate
requirement of a portfolio management system with prediction
capabilities to support the local and foreign investors to actively
engage in trading activities among different stock exchanges in
different countries. A critical study has been conducted using
supportive research papers, similar applications developed and
using various requirement elicitation techniques to determine
the functional requirements, non-functional requirements,
investor requirements, UI/UX considerations etc. The paper
further describes various technological mechanisms
implemented and system architectures used to develop the
portfolio management and stock price prediction system.
Accordingly, the implementation of Brownian Motion
algorithm-based model and LSTM (Long Short-Term Memory)
model are in detailed presented by the author. Finally,
evaluation and testing results of the completed system and stock
price prediction models are presented to prove the
successfulness of the completed application and accuracy of the
models implemented.
Keywords—stock, portfolio, prediction, LSTM, CSE
Authors —Samudith Nanayakkara, Ashen Wanniarachchi, Dushyanthi Vidanagama
SUPPORTED GLOVE PRODUCTION OPTIMIZATION WITH MACHINE LEARNING AND COMPUTER
NETWORKING APPROACH
Abstract -A supported glove, also known as a partial glove, is a
significant latex product that aims to provide safety and comfort
during industrial hazards. Therefore, the essential requirement
of supported glove manufacturing is to produce a quality
product. This research paper attempts to optimize the supported
glove manufacturing process through the machine learning (ML)
and computer system and networking (CSN) approach.
Convolution neural networks (CNNs) play a crucial role in
identifying defective gloves during the quality checking process.
Different models were tested to detect defective gloves, and
VGG16 gave the highest classification accuracy. IEEE
802.11 defines security protocols such as wired equivalent
privacy (WEP), wi-fi protected access (WPA), wi-fi protected
access2 (WPA2), and WPA2 performed robust security
mechanism. Moreover, using the production parameters such as
compound viscosity, oven and former temperatures, plant
temperature, humidity, the suggested system can predict the
defect occurrence. The predictive model has achieved the best
classification accuracy (CA) of 92% by using K Nearest
Neighbor Classifier. Furthermore, the proposed wireless
communication protocol has improved security, reliability, and
implementation using a low-cost method. The result revealed a
significant improvement in data communication by comparing
it to the other methods in terms of encryption and authentication.
Keywords—Machine learning, computer networking, convolution
neural networks, classification accuracy, wired equivalent privacy,
wi-fi protected access.
Authors —Nimshi A. Fernando, Roshima K. Hathnapitiya, Milochana G. Rathnayaka, Oshada Senaweera, Yasas C. K. Alwis, Vijani S. Piyawardana
THE FEASIBILITY OF PRE-STRESSED CONCRETE (PC) WATER RETAINING STRUCTURES OVER
CONVENTIONAL REINFORCED CONCRETE (RC) WATER TANKS
Abstract -Concrete water tanks are used to store and supply
safe drinking water and
are designed as crack-free structures. As demand for water tanks
will continue to increase in the coming years, quick construction
methods and economical design approaches will be helpful in the
selection of water tanks for relevant applications. PC will be a better
alternative for RC water tanks which are commonly used in Sri
Lanka. In this paper, design guidance for PC circular water tanks
resting on the ground is presented. Economic feasibility of both PC
and RC tanks are compared for different tank capacities. The design
and construction approaches for PC circular water tanks were
identified following BS 8110-1: 1985 and BS 8007: 1987 standards.
The Midas Gen finite element software was used to analyze the tanks.
The design outputs were converted into structural drawings and bill
of quantities. Results of the material take-offs showed that RC is
economical only for 4000 m3 or less capacity. For higher capacities
(above 4500 m3), PC tanks become cheaper by around 12-14%. The
information presented in this paper will therefore be helpful to
understand the design philosophy for the safe and economic design
of water tanks with better crack control.
Keywords—— Circular water tanks, Prestressed concrete, Reinforced
concrete, Economic feasibility, Tank capacity.
Authors —W.L.A.Chathuri Madhushani, Dr. H. D. Hidallana-Gamage
I SEE YOU : AN ANALYSIS OF THE BODY LANGUAGE OF A SPEAKER USING COMPUTER VISION
Abstract -Imagine you are standing in front of people, ready to
give a speech. You are nervous, you have butterflies in your stomach,
and you think to yourself, “I’m so not ready to do this”. But what if
you could improve your speech beforehand, without even having to
speak in front of people? This project aims on improving a major
aspect of that speech, the body language. It uses a system that
analyzes the body language of a speaker from the video of his/her
speech using computer vision techniques. It gives the speaker an
analysis of his/her facial expressions, how hand gestures were used
and how the stage was utilized during the speech. The speaker can
identify seven different types of facial expressions used in the speech
and their frequency, two types of hand gestures used in the speech
and their frequency and the percentage of each part of the stage
utilized during the speech. The speaker can also recognize the stage
transitions made during the speech. This system is also applied to
analyze the body language of the speeches of several winners of the
Toastmasters World Championship of Public Speaking. The results
of how these body language components are used by these speakers
gives the user the ability to learn from those at the pinnacle of their
public speaking journey. By providing such an analysis, this system
can greatly improve the body language skills of a speaker without
the need of human interaction.
Keywords—— computer vision, body language, public speaking, self-improvement.
Authors —Abdul Hakeem Ahmed