2020-2021
Main block
Project Title: Sound pollution monitoring and controller system
Students: Eric A. Boateng, Thompson E. Ofori
Supervisor: Gifty Osei
Abstract:
Noise pollution is a major environmental concern in Ghana today. Noise has become so common in urban areas that it is considered a blessing to have a good night's sleep at least once a week. A typical urban resident’s day begins with loud dawn broadcasts by evangelists claiming to be preaching damnation for sin, as well as loud Moslem calls for prayers over loudspeakers. You could get lucky and be away from commercial or industrial areas during the day. When you go home, there’s a blurring of music from corner stores that you have no control over. This will generally last from nightfall to daybreak, after which the religious sects will take control. This is the situation that inhabitants face on a daily basis. This increase in sound in schools, homes, and offices has proven to be a significant challenge in society. Health experts are of the opinion that excessive noise can also lead to neurosis and nervous breakdown, people may become immune to sound when they are exposed to too much noise.
The sound pollution monitoring and control system presented here consists of a sound detector, a mobile app, and a sound controller. This device detects the sound emanating from the sound source. Suppose the sound level outstretches the set threshold limit (the sound level above 48dB to 55dB). In that case, the mobile app connected to it through a WIFI module gets notified and simultaneously sends a signal to the controller to bring the sound down to a prescribed level. This device can be used in places like bars, churches, restaurants, commercial centers, and our homes to comply with the Environmental Protection Agency (EPA) policies of sound level limits. This project will help reduce the excessive noise in society and reduce the harmful effects caused by sound pollution through audio sources. The experimental results proved the functionality of the sound detector and controller in the working environment. This system provides a cost-effective means of simultaneously measuring the sound level from a sound source and automatically controlling it. This project can further be improved by using a microcontroller with more serial pins to obtain the real-time location of the device.
Keywords – Microcontroller, Liquid Crystal Display, Audio Amplifier, Cloud Server.
Project Title: Defining a propagation model for TVWS use in Accra
Students: Anthonio C. Delali, Annan Leslie
Supervisor: Gifty Osei, Godfrey A. Mills, and Frederick Abban
Abstract:
Studies on the efficient use of TV frequency spectrum have revealed and proven that TV frequency spectrum in most geographical areas is underutilized. This is noted to be very dominant in especially rural areas. This is attributed to the emergence of current telecommunication innovations including 5G and others causing the adoption of Digital Terrestrial Television Broadcast all over the world. These innovations have been implemented on a large scale in many countries and continents around the world. In many African countries, the journey of fully adopting these innovations is still in its early days. Among the recent innovations tackling the provision of broadband internet to rural areas is TVWS. TVWS is an emerging concept that was realized while studying the effective management of the TV spectrum in various regions around the World. In Ghana, the migration of TV stations to digital broadcasting in recent years has unveiled research towards the implementation of TVWS in the TV band to tackle the problems of internet access in the country on a large scale. This thesis presents the definition of a propagation model that best describes the path loss measured from the TVWS communication network per distance in Accra using the University of Ghana as a case study. In this study, A TVWS network was set up and demonstrated the use of a Geo-location Database for the assignment of operating channels to avoid interference. A management system was implemented for the regulatory body of the TVWS resource and the operation alongside the Geo-location database was demonstrated. The network performance was analyzed at various periods of time and through various channel frequency bands to understand the effect of geographical features and environmental conditions on the network. Also, signal strength measurements were performed at a number of locations in the area of study and the path loss was deduced from these measurements per distance. Finally, a proposed propagation model describing the nature of the TVWS network in the area of study is realized through statistical and comparative analysis.
Keywords - Ultra High Frequency, Television White Spaces, Geolocation Database, Protocol to access White-Space.
Project Title:A novel automated traffic control system for emergency vehicles using GPS
Students: Felbah E. Amonuah, Mills K. Amaning, Tijani Sheriff
Supervisor: Gifty Osei
Abstract:
‘Traffic’ is the term used to describe vehicles moving on a public highway. According to Wikipedia, traffic congestion is a condition in transport that is characterized by slower speeds, longer trip times, and increased vehicular queuing. Traffic congestion is a fast-rising issue that threatens the smooth running of the day-to-day activities of every individual who commutes from place to place. This traffic congestion problem may be due to a whole variety of circumstances, whether controllable or uncontrollable, and has made the need to develop new, innovative ways to control traffic and reduce the occurrence of traffic jams very pertinent, especially on the major roads in Ghana.
This project focuses on Emergency Vehicles (Ambulances, fire trucks, and police cars). Using the Okponglo Intersection as a case study, the aim of our project is to develop a system that will be able to detect when an emergency vehicle is at a distance close enough to a traffic light or intersection and adjust the traffic flow pattern by switching to a blinking red light to allow the emergency vehicle to pass freely. Thereby the response time of the Emergency Vehicles will reduce and hence more lives and properties will be saved in case of disasters.
The hardware module comprises an Arduino Uno microcontroller used for signal processing and filtering, a Wi-Fi module for communication between the Emergency Vehicle and the intersection, a GPS module embedded in the mobile phone of the driver of the vehicle, and a model of a traffic light made with LEDs. A web portal was built using the svelte JavaScript library; it was used for administrative purposes such as registering vehicles and drivers and traffic management. All the data collected are stored in a Firebase database. An android app called PREEMPT was developed for use by registered emergency vehicle drivers. The PREEMPT app was designed to provide the driver with the shortest route to an entered destination, as well as send periodic updates on location to the control station so that the appropriate actions are taken as and when needed. The individual hardware and software models performed as expected. After the integration of the hardware and the software, the fulfilled its purpose.
The system that has been developed will greatly help in reducing the losses, be it in lives or property, accrued when disasters such as fires or accidents occur. This is because it will reduce the time first-responders take to get to their destinations.
Keywords - Emergency vehicle, GPS, Preemption, Arduino UNO, Traffic light, Traffic control, Intersection.
Project Title:Smart drip irrigation system for tomato farming using wireless sensor network Students: Emmanuel G. Okyere, Suleman Sherif
Supervisor: Gifty Osei
Abstract:
Irrigation systems are becoming more important owing to the increase in human population, global warming, and food demand. This project designs and implements a drip irrigation system for tomato farming and reduces water wastage during irrigation and increases tomato yields in all seasons. Internet of things- based smart drip irrigation system using a soil moisture sensor and actuators, is designed to make water drip directly onto the roots of tomato plants when moisture level goes beyond the set threshold. Engineering system development lifecycle and waterfall model design methodologies have been deployed in the development paradigm. Using the Arduino integration development environment Proteus design suite, a working prototype of this project is designed. A Wi-Fi module has used a node to handle computations and process sensor data. It also allows remote control and view of the system using an android app. With the program embedded on the Node MCU, the DC pump can be controlled as ON- State and OFF- State depending on the soil moisture level when dispensing water unto the roots of tomato plants. This Smart drip irrigation system for deployment on tomato farms will help prevent under irrigation and over-irrigation also, help in the reduction of labor and labor cost.
Keywords - Drip Irrigation, Scheduling, IoT- Internet of Things, Wireless Sensor Network, Node MCU, Soil moisture Sensor, Automation.
Project Title:NFC contactless payment for university shuttles Students: Joseph A. Nyanu, Joshua Aryee, Michael B. Amissah, Theodoxea Kwapong
Supervisor: Gifty Osei
Abstract:
Contactless payment allows consumers to pay for goods and services using Near Field Communication (NFC) enabled cards. NFC is a short-range high-frequency form of wireless communication technology and is a subset of Radio-frequency identification (RFID) which makes use of radio waves to communicate between devices. RFID has been used for decades and has been normalized in some parts of the world for tasks such as checkout point-of-sale systems in stores, asset tracking, and even public transport payment systems. Over the years, this technology has demonstrated its ability to make transactions more secure and efficient.
Students using University Of Ghana shuttle systems are faced with concerns such as delays, money shortages due to possible lack of exact change, and during times of a pandemic, risk of infection through the transfer of physical cash. Some of these concerns apply to the shuttle conductor as well.
The goal of this project is to incorporate NFC payments in University Of Ghana shuttle systems to allow students to make shuttle payments safely and efficiently. The system consists of an NFC card that can be loaded with virtual funds, an NFC reader mounted inside shuttles that charge a student's account, an android application linked to PayStack API that allows students to check their balance and load virtual funds onto their card using MTN Mobile Money, Vodafone Cash, Airtel-Tigo Cash, and Visa/MasterCard, a server-side database application that manages all user data and transactions, and an android and web application to grant system administrators limited access to the database and allows for the creation of NFC cards for student use.
Keywords- Near Field Communication (NFC), Contactless Payment, Radio-Frequency Identification (RFID).
Project Title:Smart IoT-based hydroponics system using reinforcement learning Students: Emmanuel E. Tyron, Samuel N. Nkrumah, Israel Gayina, Joshua Nketsiah
Supervisor: Robert A. Sowah
Abstract:
Hydroponics is a method of growing plants without soil, utilizing a water-based mineral nutrient solution. The culture of raising plants through hydroponics has recently undergone a remarkable increase since the controlled atmosphere of hydroponics makes it easier to cultivate plants that are not generally produced in specific climates. Hydroponic systems need less water than soil-based systems, allow for greater nutritional control, produce healthier plant growth, and are easier to keep pest and disease-free. Because of its great efficiency, hydroponics may cover the gap in Ghana's poor agricultural productivity while also providing an ecologically benign alternative to soil cultivation. Hydroponics is essential since most farming soils have been depleted of nutrients as a result of climate change, deforestation, and soil erosion, resulting in a decline in crop production. Also, as a result of rapid urbanization, people want to grow crops in a controlled environment. Hence there is the need to develop a system that enables people to grow plants in a nutrient solution in a controlled environment. Several hydroponic systems are currently available on the market, according to internet searches. However, most items on the market are manually operated and lack automatic pH, humidity, and temperature feedback systems. Semi-automated features are costly, such as the capacity to automatically monitor and manage pH, nutrients, and temperature. In our proposed system, the system should grant the farmer a way to monitor plant growth remotely and control factors that enable plant growth. A smart IoT Based hydroponics system using reinforcement learning is a system that incorporates several technologies such as the internet of things and machine learning, specifically reinforcement learning to monitor and control a hydroponics system.
In this work, a smart hydroponics system was designed and implemented to help the user who is growing the crops to monitor, in real-time, parameters such as level of water, pH, the number of nutrients in the solution, temperature, and humidity via a web application and a mobile application. The system will control some of these parameters using reinforcement learning. This project comprises an integration of hardware and software. An Arduino UNO receives readings from the sensors sent to the Node MCU Wi-Fi module and then sent to the cloud. These readings are sent to the mobile and web applications and displayed on a live dashboard. The best parameters needed for the hydroponic crop to grow, such as the pH and the optimal water level, are fed into the Reinforcement Learning model to predict actions to control the actuators. That is done so that the system can function with little or no intervention by the farmer.
Keywords - Tensor flow, Node MCU, Open AI gym, IoT.
Project Title:Smart bin monitoring system using LoRa Students: Adjei S. Pious, Andrew K. Afful
Supervisor: Margaret A. Richardson
Abstract:
Nowadays cities are more expanded and have more population, so the amount of waste generated in the cities is increasing day by day. Waste Management (SM) is an important necessity for the environmental problem and sustainable development in many countries. One of the greatest worries with our environment has been waste management which in mixing the dustbin, the pollution of the environment is adverse. Also, poor waste management affects public health, brings many diseases, and causes poor health in the societies and the cities in Ghana. To solve the problem of poor waste management we need to focus on using smart waste monitoring and management collection, it would be more reasonable to collect them only when they are full. So, introducing a service that combines waste monitoring and waste collection to save time, money and is also essential to the health of the environment. Here, TTGO esp32, LoRa network, an ultrasonic sensor to send data from waste bins over long distances. Using a microcontroller and ultrasonic sensors we can estimate the level of garbage and make use of Lora technology. All the information will be viewed on an interface hosted on the receiver module where the users can monitor each bin. Using this interface, the user would be able to monitor the levels of fill at all times to manage the garbage collection efficiently. Finally, compared to other implementations our project utilizes the LoRa protocol which decreases the overall budget and increases the communication distance.
Keywords - Waste management, LoRa Network, TTGO ESP32, Ultrasonic sensors, Long distances.
Project Title:Fiber-To-The-Home (FTTH) technology using optical splitter channel Students: Dominic Allala Jnr., Richard N. A. Ablorh, Kwadwo Asante
Supervisor: Isaac A. Aboagye
Abstract:
The growing demand for high-speed internet is the primary driver for the new access technologies, which enable experiencing true broadband. Fiber to the Home (FTTH) networks are part of the FTT-x transmission system band in the world of telecommunications. The FTT-x family of technologies consists of a range of technologies that use optical fiber as a transmission medium to convey information. These networks which are broadband networks have the capability of transporting massive amounts of data and information at extremely high speed to the end-user. Our work aimed at implementing this technology by deploying fiber from service providers to a user’s residence. Due to the considerable economic advantages for deployment in last-mile services and optical sensing, we used the Passive Optical Networks (PON) architecture in our implementation of this technology. The design and field implementation of a secure fiber to the home (FTTH) access network serving 128 users is detailed in this work. We used (OptiSystem), a software program to design and analyze the networks. The basic components of the network are discussed, as well as the contribution of each component to the FTTH network's architecture. The architecture includes Class B security in the feeder to provide redundancy, and we develop a split configuration that separates an incident light beam from a single input fiber cable into eight individual output cables, which we call the level one splitting. In order to provide network access to all 128 end-users, each of the eight-output fiber cables is split into sixteen individual output cables. In doing this, we determine the best splitting ratio combination that will give room for fast data transmission.
Keywords - Broadband, Class B protection, FTTH, Fiber optics communication.
Project Title:Fault detection using integrated optical time-domain reflectometer (OTDR) in long haul fiber transmission
Students: Acheampong K. Danso, Akyena K. Sarfo
Supervisor: Isaac A. Aboagye
Abstract:
Fiber-optic communication has become one of the most important parts of modern communications due to the rapid development of various applications such as the Internet of things, internet data and services, medicine, and the military. They have excellent data conveying properties because of their enormous bandwidth and lower attenuation.
Due to recent developments in communication technology, there is the need for a massive amount of information to flow uninterruptedly, therefore, network outages have become unacceptable. However, optical fibers sometimes suffer faults due to some external factors that are out of our control hence, disruptions are unavoidable at some point especially in long haul fiber transmission. Some of these faults include bends, broken fiber, connectors, splice effects, cracks, and breakdowns along with an optical fiber.
This research reviews a novel technique for measuring transmission characteristics of optical fibers used in long-haul transmission. We used the Optical Time Domain Reflectometer (OTDR) device to detect faults in a fiber link using different distances and events. It is a standard technique used to investigate the quality of optical fiber installations. Therefore, faults such as a splice, break, crack, connector and other attenuation along an optical fiber can be observed by studying the visual representation on the OTDRs’ screen. We also used the Optisystem software to detect faults in long-haul transmission using different distances. The simulated results and the experimental results were compared with special emphasis on the precision of the fault location and other factors.
The results obtained show that the integrity of the signals was good and the location faults were achieved which is beneficial to help keep disruptions in the network signals to a minimum.
Keywords - Optical Time Domain Reflectometer, Fault detection, Rayleigh scattering, Events Fiber Optic Communication, Fresnel Reflections
Project Title:Outside plant (OSP) network design and route analysis for aerial and underground fiber-optic installations
Students: Aryee Shaelijah, Maximus L. Ntibrey
Supervisor: Isaac A. Aboagye
Abstract:
Our integrated world is an ever-advancing one, with data scale and complexity increasing at a rapid rate. With this in view, the need for high-performing data networking systems is on the rise than ever before. Optical fiber provides us with the needed bandwidth and speed to achieve this. However, the implementation of these installations is a very challenging task in itself. The development of other servicing channels such as water pipelines and electric cables means that the planning and design of these optic pathways are to be tactful and precise. This has not been the case, more so in the country. Unmethodical planting of fiber cables spells out great consequences, contributing towards inhibiting the steady development of a nation. Naturally, measures have been placed to curb this but mismanagement of funds and inadequate public broadcast has made results unyielding.
This project presents an investigation into the route design and analysis of fiber architectures, taking into account aerial and underground installments. This involved the use of scientifically appropriate procedures, along with implementation, examination, and gathering of results for validation. In our project, we realized to an extent, optimum structural design and administration of the backend framework of fiber systems. A high-level design was produced, following technical site surveying to which a low-level design is produced. Scientifically-based procedures and operations were further suggested in specific areas, with an estimation of parameters, gathering of resources, and implementation. Testing and analyses were made, following adjustments where necessary.
In the end, losses of 224.84 to 22.04 dB from distances 0.15 km to 5 km were observed. Individual components in the OSP architecture contributed to fixed losses of 0.7, 10.5, and 10.6 dB in the OLT, FDT, and FAT respectively. Thus, actual losses from cable length and nature of route ranged 1.04 dB to 2.24 dB. A slope of 0.24 dB/km was calculated and this meets the required path loss of less than 1 dB/km in fiber optic transmissions.
Keywords - Outside Plant, High-Level Design, Low-Level Design
Project Title:Detection and Management of Rheumatic Disorders using Machine Learning and Blockchain
Students: Mohammed A. J. Kassim, Yiwere Aminu
Supervisor: Godfrey A. Mills
Abstract:
Rheumatic disorders have been identified by World Health Organization to be among the leading causes of acute disability in adults. These disorders comprise over 200 diseases, the majority of which are autoimmune. The autoimmune rheumatic disorders cause the body’s immune system to attack healthy cells around the joints, muscles, bones, and vital body organs, resulting in an inflammation of that region. Other forms of rheumatic disorders result in a breakdown of bone and cartilage in joints rather than an inflammation. Early diagnosis and treatment can slow the progression of many rheumatic disorders and lead to remission if properly managed. Diagnosis of rheumatic disorders for clinical decision is a complex process that requires information such as patient history, physical examination by a rheumatologist, and laboratory tests. The goal of this project is to leverage a machine-learning-based decision support system to help with the fast screening of patients for early detection and management of the three most common rheumatic disorders (Rheumatoid arthritis, Osteoarthritis, Systemic Lupus Erythematosus). A labeled dataset of 100,000 instances was generated to train and test the machine-learning model using standards by the European League Against Rheumatism (EULAR) and knowledge from specialists. The dataset was further decomposed and grouped by age and gender. A feedforward neural network architecture comprising 45 input neurons, 2 hidden layers with 10 and 15 neurons on the first and second layers respectively, and 4 output neurons was established. The network was trained using 70% of the dataset and 30% was used for testing and validation. The model was tested for the prediction of the three disorders. Results showed detection accuracy of 97.48%, an average sensitivity of 96.80%, and an average specificity of 97.50%. Both clinical tests and blind tests were carried out by the rheumatologist and results showed detection accuracy of 97.01%, with average specificity and sensitivity values of 100%. A web application was implemented for the model to aid clinicians in the fast screening process. Block-chain technology was introduced in the application as a mechanism to secure patient data. The system developed will be able to address the problem associated with the scarcity of rheumatologists in the country. It will serve as a tool for quick screening of patients, decentralize expert knowledge to lower-level medical practitioners, and as a resource for training health personnel on the diagnosis of rheumatic disorders. This will help prevent life-threatening complications that may arise due to late diagnosis of rheumatic conditions.
Keywords – Rheumatic disorders, Autoimmune, Inflammatory, Machine learning, Blockchain.
Project Title:Optimization techniques for monitoring and control of microgrids with distributed generation networks
Students: Baron Afutu, Dorothy Amarh, Eugene Animante
Supervisor: Godfrey A. Mills and Godfred Mensah
Abstract:
Microgrids (MG) are small-scale power grids that can work independently or in conjunction with the main electrical grid in a given area. They usually contain renewable energy sources (RES) such as wind turbines and solar panels, along with energy storage units. MGs have become an area of interest to utilities because are environmentally friendly and offer cost-effective, clean, and stable electricity operations. Nevertheless, the combination of different RESs, considering their intermittent nature, pose several challenges. These include voltage synchronization, distortion in current harmonics, over/under voltage, frequency deviation, transient voltages, power supply, and load balancing, optimal source scheduling considering source timing and ramping rate, island detection, and anti-islanding, and finally, effective monitoring, control, and coordination. This project introduces some optimization methods for the development of a scheduling algorithm and a control system in an attempt to solve some of these problems. An optimization technique that is based on both Linear Programming (LP) and an Artificial Intelligence (AI) based solution was evaluated for implementation. The AI-based solution was found most suitable for the adoption and this was implemented in the proposed management of the operations of MGs. Numerical experiments were employed to test the proposed model at 11kV and this was followed by the development of a prototype using lower voltage levels. Using the University of Ghana load profile as a case study, the optimization model was tested using four sources, a solar photovoltaic source, an energy storage system, a diesel generator source, and the main grid.
Results show that the system was able to effectively optimize the scheduling operation of the proposed MG network. The implemented model resulted in a 23.45% reduction in the total cost of the operation of the MG at an average time of 13.20s per iteration. The system was able to achieve the set requirements by IEEE 1547.2, 2030.7, and 458 standards. This solution will be very suitable in helping utilities running MG to be able to manage effectively the operations in the network to solve the challenges of over/under voltage, frequency deviation, power supply, and load balancing, optimal source scheduling considering source timing and ramping rate and island detection and anti-islanding for effective control and coordination of the microgrid.
Keywords - Microgrid, Distributed generation, Renewable energy sources, Optimization, Linear programming, Genetic algorithm, Source scheduling, Anti-islanding.
Project Title:Detection of land encroachment within transmission line corridor using convolutional neural networks
Students: Djarbeng Richard, Gyamera B. Collins
Supervisor: Godfrey A. Mills
Abstract:
Overhead electricity transmission lines have a corridor which is supposed to be free of any encroachment, to make management of the transmission lines very easy and also to protect human life in the event of any line drop or casualty. In the urban areas where the demand for land is quite high, people often encroach on this corridor, making the maintenance work of the transmission utility quite difficult. Transmission utilities are therefore faced with the challenge of monitoring these corridors. Conventional methods for monitoring the transmission line corridors include occasional visits to the site, using vehicles to drive through the length of the corridor. Technologies such as drones or helicopters have also been used. These methods when used for periodic monitoring are time-consuming and expensive in relation to the coverage area. This project work provides a monitoring system for the identification and real-time detection of unauthorized structures such as shops, housing, etc. in a transmission corridor using satellite imagery and machine learning technologies. The detection system is driven by convolutional neural network architecture. The network was trained using multiple objects along sample transmission lines. 3600 objects comprising of buildings, shops, and lands were used to train and test the network. The system was tested using the transmission line corridor of the Ghana Grid Company (GridCo) from their Volta station at Tema to the East Legon Bulk Supply station to the Achimota BSP. The pilot corridor involves 50 transmission towers with variable span lengths. The model was able to detect an accuracy of 97.50%. In order to track changes in structural developments on the transmission corridor, the model was integrated with a visualization tracking system. Satellite images from Sentinel-2 satellites using Google Earth Engine were used. Through this solution, the transmission line utility will be able to remotely monitor their transmission lines in real-time using a web-based application to avoid any threatening situation before it occurs.
Keywords – Convolutional Neural Network, Encroachment, Transmission Corridor.
Project Title:Action recognition in surveillance cameras
Students: Isaac F. Dickson, Kelvin G. Yahaya
Supervisor: Longdon N. Sowah
Abstract:
The ability to detect the actions which occur in a video is essential for automatic understanding and post-action security analysis. This reduces greatly the cost of human resources for smart surveillance. A lot of the methods designed for intelligent surveillance focus on the detection of single events by just an individual in a well-segmented video rather than on multiple individuals performing different actions at the same time which reflects the conditions of the real world. More decorative of an intelligent surveillance system is the ability to detect the actions of multiple people at the same time in real-time. Over the past years, researchers have investigated techniques that can automatically recognize human activities.
This research field is commonly known as Human Activity Recognition (HAR). HAR includes many steps which include signal or data acquisition, pre-processing, data segmentation, feature extraction, and classification. There are different techniques that have been tested and different solutions have been proposed to achieve classification results. Based on the requirements and specifications, different solutions are chosen based on their strengths and weaknesses. In this paper, we present a method of recognizing human actions using pose estimation. In our project, the recognition of human action is based on human skeleton data. We used an open-source algorithm, OpenPose, to detect human skeletons (joint positions) from each video frame and utilized the skeleton as raw data to extract features and make classification by using machine learning algorithms. There are several methods for action recognition such as 3D Convolution Neural Network (3D-CNN) and Long Short Term Memory (LSTM) to directly recognize actions from video. However, it takes a lot of time and it is difficult to train the large neural network, and also lacks clear understanding. On the contrary, the features of the human skeleton are straight to the point, intuitive, and easy for differentiating different human actions.
Thus we chose the human skeleton as the base feature for action recognition. The human skeleton data approach extracts the human skeleton features using 2D positions of human skeletal joints through an OpenPose library. The approach was trained and tested on publicly available datasets and machine learning techniques were implemented for recognizing nine activity classes including stand, walk, run, jump, sit, squat, kick, punch and wave. Our system was tested and it was able to extract the skeletal data of the individual and classify their actions. The main recommendation for this project is to train and test the system on more human actions to enable it to classify more actions. Also, another recommendation is to incorporate the system which is an actual video surveillance camera, and also to introduce an alert system to alert persons of interest in case a harmful action is detected. An example is to alert an administrator when the system detects a harmful action during crowd surveillance.
Keywords - Human Activity Recognition (HAR), Feature extraction, 3D Convolution Neural Network (3D-CNN), Long Short Term Memory (LSTM).
Project Title:Automated timetable generator application
Students: Sydney Dapilah, Robert Ayetey
Supervisor: Longdon N. Sowah
Abstract:
This is a project on how constrained satisfaction artificial intelligence is used in the design and building of an automated timetable generator. The application mainly seeks to solve the issues of delayed and error-free timetables. Scheduling course timetables for a large number of courses for departments can be a difficult task. It must be done manually by a designated staff despite the fact that timetables generated are frequently inaccurate and take up a lot of time to complete. Timetabling is a highly constrained combinatorial problem and this project makes use of the effectiveness of constraint satisfaction artificial intelligence algorithms in solving constrained automated problems. In artificial intelligence, constraint satisfaction is the process of finding a solution to a set of constraints that impose conditions the variables must satisfy. A solution is therefore a set of values for the variables that satisfy all constraints. In the above problem, the basic constraints to be satisfied are, only one course can be scheduled at a venue and within a certain period of time, a student can attend only one lecture at a particular time, and a lecturer can only teach one course at a particular time. The idea here is first to have a metric that tells if a timetable is optimal or not optimal and secondly to use this metric as a guide to generating a timetable that satisfies all the basic constraints. When compared to using genetic algorithm methods that deal with creating random timetables until an optimal solution is found, constraint satisfaction methods of creating timetables are very fast because the timetable is scheduled knowing what a good timetable should be and there is control over the outcome of a generated timetable. The genetic algorithm process may take forever to complete.
This project demonstrates the power of intelligent modules in automating tedious tasks. While the current version of the project only considers basic constraints, more constraints can be added to suit the minor needs of timetable users and the application can be scaled up to be used across the entire university.
Keywords - Constraint Satisfaction Algorithm, Artificial Intelligence, Genetic Algorithm, Timetabling.
Project Title:An IoT-enabled smart door for monitoring body temperature and face mask detection.
Students: Titus A. Adotei, Samuel N. Owusu, Emmanuel O. Boadu
Supervisor: Longdon N. Sowah
Abstract:
The outbreak of the deadly COVID-19 virus has been a big worry to the world, and as a result, there is the need to take precautions at every place. Because of the COVID-19 pandemic, wearing a mask is required in broad daylight spaces, as appropriately wearing a mask offers the greatest preventive impact against viral transmission. Body temperature has likewise become a crucial parameter in determining the health status of an individual. In public places where there are lots of crowds, it is very difficult to manually monitor and ensure every individual is abiding by the COVID 19 protocols. In this work, an IoT-enabled smart door utilizes a machine learning model for checking body temperature and face mask detection. This proposed model can be utilized for any shopping center, lodging, airports, banks, apartment entrances, and so forth. As a result, a cost-effective and reliable method of utilizing computer vision and sensors to build a healthy environment. The project comprises of integration of hardware and software. The proposed system uses contactless infrared temperature sensors to measure the temperature of the individual at the entrance of public places. Our system also implements real-time deep learning models for face mask detection and classification. The module has three classification instances which include; individuals who wear face masks properly, improperly and without a face mask using MobileNetV2, CNN, and Tensor flow libraries using the transfer learning approach. Access is granted to the individual if the output from the temperature measurement module and the face mask detection module meets the acceptable threshold that is predetermined by the system. Also, the system is interfaced with a desktop application that keeps records of individuals for further control and monitoring.
Keywords - Machine Learning, Computer Vision, Tensorflow, MobileNetV2, Transfer Learning, Raspberry Pi, Covid-19, Open CV, Python, PyQt5
Project Title:Enforcing social distancing using Bluetooth and computer vision
Students: Cephas N. A. Hammond, Emmanuel A. N. Adjei, Ahornam Agbotse
Supervisor: Longdon N. Sowah
Abstract:
Since the onset of the COVID-19 pandemic, there has been a great shift in normal activities across the world. According to the World Health Organization (WHO), COVID-19 is spread between infected people through close contact and the most effective way to reduce the transmission of the virus is social distancing. This project seeks to design and develop innovative technological approaches through Bluetooth technology and computer vision to ensure compliance with the social distancing protocol.
A mobile application was designed that uses Bluetooth technology to track the distance between the user and surrounding devices within the range of 1 meter. The mobile application scans for any Bluetooth device nearby and upon detection, calculates the distance between the user's phone and the identified device. Once the device is near the user of the application, the application alerts the user in the form of vibrations and sound. Testing of the designed mobile application shows that the application can detect devices within the 1-meter range and carry out the necessary alerts as specified by the user.
A computer vision system was also developed to monitor the social distance compliance level among people. This system leverages the OpenCV library and Deep learning based on YOLO training datasets. Data frames from a camera feed are fed into the system and the system identifies people in the frames and performs a pairwise distance estimation. Upon violation detection, the images are sent to an administrator to view the violations that occurred. Testing shows that the system has a high accuracy of detecting people and performing distance estimation between pairs. It also shows that the images and violations are successfully sent to the administrator for monitoring.
The performance of the proposed solutions shows that by leveraging technology, social distancing can be adhered to, reducing the transmission of COVID-19.
Keywords - Bluetooth, social distancing, computer vision, distance estimation, object detection.
Project Title:Smart waste management system for residential areas
Students: Mortey K. Emmanuel, Maxwell E. Acquah
Supervisor: Longdon N. Sowah
Abstract:
In the current day situation, we mostly realize the overflow of waste in bins and trash cans at certain unexpected places especially residential areas. Waste continues to spill over day by day in these places because of the higher rate of expansion of residential areas. Our population is definitely proportional to the amount of waste generated and this accounts for the rapid generation of waste in residential areas.
Based on research conducted by Quartz Africa, there has been an indication that, while waste management is a widespread issue in Ghana, Accra seems to be the most affected city creating from about 2,000 to 3,000 metric tons of waste within 24 hours. In 2012, the World Bank evaluated that the fall in Ghana’s economy was due to poor sanitation.
Given this, we intend to propose a unified system of both software and hardware that is based on the concept of eliminating all these issues by enabling anyone to dispose of their trash at any point of time at their convenience. The software i.e., the mobile app would have a customer and a driver aspect where the customer can schedule or order for immediate garbage pickup at any time. It would make use of Global Positioning System (GPS) to enable real-time tracking of the location of the connected pickup driver.
Here, the customer would be connected to a nearby user who has already signed up as a driver on the app and is currently online. One can also have a real-time feed of the driver’s location as he approaches the customer. Another phase of this project aims at aiding individuals to keep track of the garbage levels in their cans and bins by introducing an IoT hardware system. The device consists of an ultrasonic sensor, an Arduino UNO, and a Wi-Fi module that would communicate to servers and give feedback to the mobile app. Fairly, everything that happens on the hardware phase is subsequently shown in the software thus, the percentage of rubbish in the bin is displayed in the mobile app.
This is what our project seeks to do to curb the problem and to ensure a safe and healthy environment.
Keywords - Global Positioning System, waste management system, IoT.
Project Title:Classification and detection of labour types
Students: Agyemang Ernest, Frimpong M. Warren
Supervisor: Godfrey A. Mills and Elsie E. Kaufman
Abstract:
A research exercise at various resource-limited hospitals in Ghana revealed that a conventional method of monitoring uterine contractions is employed. This method is time-consuming and ineffective with a likelihood of misrepresenting data on uterine contractions. Uterine contraction (UC) is an important clinical indicator for monitoring uterine activity. During pregnancy, early detection of uterine contractions leading to preterm delivery is crucial for preventing early birth. Constant monitoring of the fetus could help to detect early symptoms and anomalies that can be a sign of premature childbirth and other fetal complications that could be of major concern. During the later stages of pregnancy, women go through labor to be able to give birth. One way of observing labor progress is through the monitoring of uterine activities. Uterine contractions become more frequent, last longer, and are more intense as labor progresses. Health personals find it difficult to monitor and track the progress of labor. There is therefore a need for a system that can potentially overcome the identified challenges.
In this project, the proof of concept for the development of an automated uterine contraction monitoring system designed for use. This project seeks to provide an automated system that will monitor and classify uterine contractions and notify the health personnel when there are some irregularities with limited human intervention. The proposed system developed provides a way to monitor uterine contraction during labor. The system records and computes the intensities of the uterine contractions. The recorded data is sent to another system on the desk of the health personnel through wireless Bluetooth communication. The system classifies records into regular and irregular contractions and alerts the health personnel when necessary.
The developed system was tested and was able to determine parameters that are necessary to monitor the progress of labor. The system is able to classify recorded uterine contraction results (intensities) and also determine whether contractions are regular or irregular. This project will also be useful for doctors, nurses, and other health personals in managing their patients. The system developed will provide enormous benefits to users in the monitoring of uterine contraction and the associated risk and complications during labor.
Keywords - Uterine contraction, Monitoring, Labor, Pregnancy Detection.
Project Title:IoT based traffic controller surveillance system
Students: Mawuko Tettey, Jude O. Sackey, Cyril K. Acquah
Supervisor: Wiafe O. Banahene
Abstract:
Destruction of public amenities has been a major problem in Ghana since many of these amenities are not being monitored. One of such amenities is the traffic lights. Until recent years, city authorities focused on the detection of electrical faults and damages to traffic lights and associated equipment. However, this situation has changed after the popularization of the use of solar panels to power this equipment. This new development has made traffic controllers valuable assets since the cost of installing solar-powered traffic lights ranges between GHS100,000.00 to GHS200,000.00.
Traffic controllers have now become a target for thieves who steal this equipment and sell it as parts for their personal gains. The damages further contribute to traffic congestion and unsafe roads since the traffic lights would not be working properly to control the flow of traffic. The congestion is mainly felt at intersections where a deadlock usually occurs when the traffic light controllers are damaged, faulty, or off as a result of a power outage. There is therefore the need to monitor these controllers at the various intersections to help report any fault or damage as quickly as possible so a response team can be sent immediately.
In this project, an Ultrasonic sensor and PIR sensor were used to detect the presence of any approaching person. Once the person gets close to the mesh protecting the controller, a warning is given through an alarm system with a flashlight to hopefully ward off the intruder. If the unauthorized person continues to breach the fence, video footage would be taken through the camera and where possible identify the person through facial recognition software. An alert would be immediately sent through the web application showing which controller is under attack.
In the future, the controllers could be designed to be fitted underground to help minimize the damage that can be caused to them. Also, with the growth of drone technology, self-deployment drones could be commissioned to respond to such threats when detected since they would take less time to reach that area. This would go a long way to protect the controllers, save cost and also reduce traffic congestions in the long run.
Keywords - Traffic Controller, DeepFace, Surveillance, Image Processing, Face recognition.
Project Title:IoT based smart home surveillance security system
Students: Jeffrey E. O. Hwedieh, Edmund B. Essah, Keziah A. Odoi
Supervisor: Wiafe O. Banahene
Abstract:
The transfer of data from household devices over the internet has been made by the introduction of the Internet of Things (IoT). IoT can be viewed as a large network that allows users to connect their devices and share data over the Internet. Using IoT devices has the advantage of allowing a user to be able to control the device wherever they are. The importance of integrating IoT in security systems can be driven off from the stated advantage. In the day and age of people becoming more security conscious, it is important for people to be able to have security systems installed in their homes and institutions, without bearing the exuberant costs, and also being simpler to operate. Homeowners and institutions resort to using CCTV cameras for surveillance but they perform the function without the added feature of providing security. This problem is what this project seeks to solve. By incorporating the element of facial detection into the camera surveillance system, we hope to provide a robust security system that notifies owners when intruders are present on their estate. By doing this we hope to provide people with the security they need, a security system that is accessible from any place at any time, a system with a simpler mode of operation, all for a cheaper price. This project aims to develop an IoT-based smart home surveillance security system that is smart enough to detect the presence of a flame, detect motion, detect smoke, read the temperature of the home and also detect possible LPG leakages. The system also features face detection that can capture the faces of intruders in images.
Keywords - Closed-circuit Television, General Purpose Input Output, Open-Source Computer Vision, Web Server Gateway Interface, Integrated Development, and Learning Environment.
Project Title:3D-Based navigation system for the University of Ghana using extended reality technology
Students: Joseph K. Asante, Petra N.A Sarpong, Bintu Alhassan
Supervisor: Wiafe O. Banahene
Abstract:
Traditional or 2D navigation systems are useful tools when it comes to finding your way around new environments. These systems are helpful in locating places or buildings of interest when the user has an idea of how these buildings look like.
Such maps are not very effective as they represent real-world locations using vector graphics such as line paths to represent roads and marker or pin images or text to represent buildings. During Global Positioning System (GPS) outages, internet lags, or coordinate data anomalies within these systems, the users could be misdirected and the building in search of might be at the wrong position in relation to the real world. This project seeks to design and develop a 3D navigation system to visualize real-world locations within the University of Ghana, hence making it easier to navigate around such locations. The system provides a 3D real-world simulation for the user to detect unlabelled buildings of interest to navigate much easier through a location even without being physically present. This is done using a Virtual Reality subsystem and detailed 3D models. The system also gives information on specific buildings by allowing the user to scan the building using his or her phone’s camera and then augment useful data about the building upon detection on the screen of the user. This is done by an Augmented Reality subsystem. These two subsystems, that is, Virtual Reality (VR) and Augmented Reality (AR) are what are collectively known as Extended Reality systems commonly known as XR systems. The developed system was tested by allowing people to use the system to find certain locations and the results show that the system was a better navigation tool as compared to the 2D or traditional navigation systems. This system will be especially beneficial to anyone visiting the university for the first time. This includes tourists and first-year students who will need help detecting important landmarks such as lecture blocks and halls for first-year students.
Keywords - Virtual Reality (VR), Augmented Reality (AR), Extended Reality (XR), Navigation, Global Positioning System (GPS), Three Dimensional (3D), Two Dimensional (2D).
Project Title:Farm animal management system for poultry and livestock using artificial intelligence
Students: Sunday Binakin
Supervisor: Robert A. Sowah
Abstract:
Artificial intelligence has evolved to the point that it has revolutionized every area of the global economy, including agriculture, during the previous several decades. The agricultural industry's traditional approach is experiencing substantial change. As a result of the growing demand for enhanced animal products, AI has been developed as a powerful tool to aid farmers in monitoring and diagnosing agricultural diseases. Furthermore, using AI, farmers would be able to detect sick animals at an early stage. Because traditional animal sickness diagnosis needs expertise and takes a long time to analyze, AI is used in conjunction with image processing to provide accurate, rapid, efficient, and low-cost results.
In this study, a drone was used to capture images in real-time, which were then sent to a cloud server for the classification of various illnesses affecting farm animals. A mobile and web app system was created to automate illness diagnosis, making it quicker, easier, and more accurate. This application will assist farmers or users in taking and examining photos of farm animals in real-time, making disease diagnosis easier.
The system was built using a variety of datasets from illnesses in both poultry and cattle, including salmonella, Newcastle disease, bluetongue, coccidiosis, bluetongue, lumpy skin disease, and bovine viral diarrhea (BVD). The results demonstrated that the system was capable of correctly predicting farm animal diseases. The application will benefit farmers, particularly those who operate on a large scale.
This system will also assist farmers in monitoring the total number of animals on their farm as well as the diseases associated with each group of animals. In general, this project will improve animal welfare while also assisting farmers financially.
Keywords - Artificial intelligence, real-time imaging, Convolutional neural network, Machine learning.
Project Title:Cash crop disease detection using machine learning and unmanned aerial vehicle
Students: Yussif I. Gyanko, Awuah Y. Ransford, Owiredu Francisca
Supervisor: Robert A. Sowah
Abstract:
Agriculture being the stronghold of Ghana’s economy is being gradually threatened by diseases causing significant losses to farmers and threatening the economy of the country. Cash crops in general are resourceful and are the boosters of the country’s economy, due to this we based our project on detecting diseases that affect cash crops such as cocoa; black pod disease, frosty pod disease, and healthy pods of cocoa. Coffee; Cercospora, rust, phora, and miner whiles cotton was classified as to whether the plant or leaf was diseased or healthy. Farmers in Ghana face a lot of issues in the sector due to the fact that they are not able to detect the diseases in their early stages, they also are not able to sometimes determine the type of disease affecting the crop hence the urgency to reduce the spread would be lost leading to the destruction of crop yields and waste of manpower. Another issue worth discussing is the fact that they import experts to detect the disease physically or manually, this is not only a waste of resources but also risky since mistakes could be done. A lot of researchers have researched the topic and a lot has been done to solve it. Some of which involved building an image processing system, using cameras to take pictures, and through computer vision tried to solve it in which all had their disadvantages and we emphatically listed all their flaws in our reviews. We introduced a new approach based on convolutional neural network and machine learning with the help of drones serving as carriers to our hardware (raspberry pi camera and raspberry pi model 4) which takes real-time images and sends them to the server for instant prediction. The goal of this project is to detect, identify and accurately predict the type of disease affecting the cash crop. The proposed system is a very intelligent and accurate method in the detection of plant disease which will in a result minimize the losses and increase our economic benefit.
Keywords - Convolutional neural network, Plant Disease detection Unmanned Aerial vehicle, Machine learning, Artificial intelligence.
Project Title:University of Ghana smart shuttle system with partial COVID-19 protocols
Students: Bryan Vukania, Sola J. Eniafe, Henry N. Tandoh
Supervisor:Wiafe O. Banahene
Abstract:
A smart shuttle system is a system that incorporates several technologies such as real-time tracking of vehicles as well as other services such as providing the estimated time of arrival of the shuttle to a particular location and a means for users to pay for their travels. In this work, a smart shuttle system was designed and implemented to help the University of Ghana students to book a seat, locate, in real-time, the position, and estimate the arrival time of shuttles via a web application and a mobile application. The system will also include a payment system to provide a means for students to pay for their rides without the use of physical cash. This project comprises an integration of hardware and software. A hardware system with a motion-triggered hand sanitizer dispenser. The mobile app has a feature that allows students to pay for a ride in the shuttle through its payment system. The hardware system is placed in a shuttle and it allows users to scan NFC cards and transmits via an Application Programming Interface (API) endpoint to the backend of the web application which sends the data to the firebase to indicate that the user has boarded the shuttle, it also has components such as buzzers to alert driver on a pending exit or entrance of a passenger when the shuttle is approaching an exit or pick up location. When a user signs in or signs up on the mobile application he or she is redirected to the map page where he or she can see the location of the bus on Google Maps as well as the estimated time to the respective pickup bus stop. When the user wants to pay for a ride, there is a payment page on the web application where he or she can pay with either mobile money or credit card. Our system was tested for the various functionalities mentioned above to make sure that the system was fully operational. As such, the primary objective of this project is to allow students to be able to locate the position of the shuttle on campus at any instant in time as well as the provision of a payment system for students to be able to pay for using shuttle services was achieved. The main recommendation for this work is the addition of more features to make it more accessible for people to be able to use it efficiently.
Keywords - Flutter, NFC, Real-Time Tracking
Project Title:Program accreditation management system for the University of Ghana
Students: Laureen I. Zormelo
Supervisor:Godfrey Mills
Abstract:
Accreditation of Programs is an essential aspect of educational systems globally. It provides insurance that the program's curriculum and quality have been evaluated by a centralized body to meet the standards of a particular profession. Accreditation thus ensures that one is eligible to work in a particular profession or further education.
Accreditation is not done once but rather periodically hence the likelihood of programs not being reaccredited when the next accreditation date is due. This is most likely to be caused by human error which can lead to shutting down of the program and the centralized body invoking penalty.
The system developed provides a web application that monitors and tracks the accreditation status of programs run by the University of Ghana. It was implemented using Microservice Architecture and Model View Controller developmental pattern and tested with pseudo data to test the functionality of the system and to obtain results. The system met all requirements set and it proves to be of great benefit to both the society and the institution by ensuring that programs are eligible to be run by the institution at all times and also ensuring that the quality of programs is kept to standard.
Keyword - Accreditation, Microservice Architecture, User Interface Development, Web Interface
Project Title:Fingerprint Based Biometric Attendance System
Students: Nana O. Frimpong, Enoch J. O. Afari, Clement B. Ankomah
Supervisor:Percy Okae
Abstract:
This project is focused on the development of a biometric attendance system to ensure that taking attendance in the Computer Engineering Department of the University of Ghana is much more efficient and secure. The manual system of taking attendance is done by the use of pen and paper, where students write their names, index numbers, and sign on the sheet of paper. This makes the system unreliable because students can write names for friends who are not in class. Also, the ability to compute the attendance percentage becomes a major task as manual computation produces errors, and also wastes a lot of time. For the stated reasons, an efficient attendance management system using biometrics is designed. The biometric attendance system was achieved using a fingerprint device and the recorded attendance was stored in a CSV file. During the process, it was observed that while the system has a high rate of accuracy, there is also a possibility of false negatives in verification. However, the rate of these is negligible. A device was developed for recognition and verification of fingerprints that were capable of uploading records to a CSV file, thus, improving the reliability and security of the attendance-taking process. This project can be applied in many institutions apart from schools such as corporate institutions and hospitals. This project proved to be a more viable, modern option to the traditional pen and paper system and in the future, can be improved upon by implementing an IoT solution to allow the attendance records to be accessed online.
Keywords - Attendance, Biometric, Fingerprint.
Project Title:Automated handwashing system using Arduino
Students: Felix Y. Tamakloe
Supervisor:Percy Okae
Abstract:
As technology continues to evolve, systems need to be made that substitute human effort and intelligence for mechanical, electrical, or computerized action. These automated systems serve to improve quality of life and minimize the amount of human input required, and such technological advancements need to be applied to some of the most common activities such as handwashing. This is because manual handwashing is characterized by wastage of water and can prove ineffective if the user is not disciplined.
An automated handwashing timer device is implemented for assisting a user in washing their hands for a proper amount of time, for example per the World Health Organization guidelines. The control board is used to sequence the hand washing process, with each step automated to proceed as required. This system is powered by Atmega328 in an Arduino UNO microcontroller. The timer system also includes a sensor, a display, pumps, and an alert system.
When a hand was detected, the system activated the soap dispenser for a given period to enable the user to apply soap, then activated the water dispenser to enable the user to rinse hands. The LCD, buzzer, and LED were used to guide the user through the various stages of the process.
Compared to other implementations, this approach made provision for both soap and water to be dispensed thus automating the whole process.
The device was found to dispense both soap and water as required with the appropriate audio-visual aids guiding the process. I recommend that future versions of this project be implemented using a solenoid valve instead of pumps. Also, a much more sophisticated sensor should be used to ensure that the system is activated only when a hand is detected and is in a position to receive soap or water. Finally, a real-time clock can be integrated into the system to ensure that more detailed handwashing information can be stored and exported.
Keywords - Cross-Contamination, Ultrasonic, Pathogens, Pandemic, Audio-Visual, and Automated.
Project Title:Integrated solar tracker using Arduino
Students: Osman M. Raji, Jude Lade
Supervisor:Percy Okae
Abstract:
Solar energy is fast gaining traction as a viable source of renewable energy. Solar heating, photovoltaic, solar thermal energy, solar architecture, molten salt power plants, and artificial photosynthesis are all examples of ever-evolving technologies that harness the Sun's radiant light and heat. Solar trackers aim for solar panels (PV panels) or modules directly at the sun. In this paper, a solar tracking system using Arduino is built. A sun tracking system is created with Arduino in this study.
This system captures free solar energy, stores it in a battery, and then transforms it to the appropriate alternating current. It allows the energy to be used as a standalone power source in ordinary households. This system is built to adapt to its surroundings as quickly as possible. Any software and hardware issues will be minimized or removed. Our system is put through its paces in terms of real-time responsiveness, reliability, stability, and security.
Weather, temperature, and mild mechanical stresses are all factors that our system is built to withstand. The solar tracking system is the most effective technique to improve the efficiency of solar panels by tracking and following the sun's movement. With the help of this system, solar panels can improve the way of sunlight detection so that more electricity can be collected as solar panels can maintain a sunny position.
Thus, the project discusses the development of dual-axis solar-tracking developers using Arduino Uno as the main controller of the system.
This project seeks to maximize the output power from the sun by the automation of the movement of solar panels to properly align at the sun’s maximum light intensity.
Keywords - PV panels, Arduino UNO, dual-axis.
Project Title:Waste monitoring using IoT and payment platforms for users in domestic homes
Students: Boateng A. Aaron, Dzegblor Mawuko, Sadat Alhassan, Kale Samuel
Supervisor:Percy Okae
Abstract:
The increase in human population and urbanization in our world leads to the increase in the amount of solid waste produced because humans tend to utilize several products and materials within a short time thereby creating a large amount of waste. Although these solid wastes are kept in a garbage bin, the monitoring and collection of the waste are not effective. Also, the payment of waste bills is done manually where the vendors have to move from door to door to collect monies, and the records are stored on monthly record cards. For this work, a real-time monitoring system using the Internet of Things (IoT) was developed. In conjunction with the concept of using the Internet of Things (IoT), sensor technologies are used to measure the level of the waste in the garbage bin and Wi-Fi Technology to connect to the internet and transmit the data (measured values from the sensor) in a cloud database. The data from the cloud storage can be connected with the waste management company server to monitor every individual or customer’s waste bin. The customers also have the opportunity to view and monitor the status of their garbage bin and make appropriate payments for their waste bills when the time is due. Furthermore, solar cells can be incorporated to provide a constant supply of power, multiple sensors for more accurate readings, and a Global Positioning System (GPS) for the location of garbage bins. This project will help waste management companies to track payments of clients without being deceived and also curb the overflow of garbage bins.
Keywords - Waste Monitoring, IoT, Global Positioning System.
Project Title:IoT-based agricultural system
Students: Dennis E.Quansah , Daafuor O. Leslie, Grace N. Yowome
Supervisor:Stephen K. Armoo
Abstract:
In modern times there has been a rapid growth in the use of technology and the “internet of things” (IoT) based systems. Technology and IOT systems have contributed to the standard of living by increasing efficiency and productivity in the use of resources as a way of making life and work easier.
Agriculture in Ghana has been essential to the development of the country. However, its contribution to the economy slowed down for the past four years according to a record presented by the Ghana Statistical Service in the year, 2018. This has occurred due to the irregularity of the weather and climate over the past years leading to soil depletion, as well as the inability to embrace the usage of modernization in agriculture. Another problem that the agricultural sector has faced is the failure of gardeners or farmers to accurately apply the right amount of nutrients to the soil. These factors have consequently led to adverse effects on food production and crop yield.
It is however necessary to help grow our agriculture sector in the country. The growth in agriculture will provide more food for the Ghanaians citizens which will curb the increase in inflation and cost of food products, revenue and exports will also be generated for the country, thus bringing in foreign exchange. Hence, an efficacious technique for growing the agriculture sector is the use of the IoT-based Agricultural Monitoring System.
This makes it easier for farmers to require information about their plants. It also encourages less labor with more efficiency since irrigation and fertilizer application are automatically triggered or halted. Consequently, this will help in bountiful harvests, decrease the cost of food products as well as grow the agricultural sector through exports. The system is applicable in both large-scale and small-scale farming.
Keywords - IoT Based Agricultural Systems, Sensors, NodeMCU, Arduino Uno, ThingSpeak server, POST requests, Fertilizer Pump, Water Pump.
Project Title:Integration and use of OCR technology in an automated speed detection system for security scanning and database tracking
Students: Yarboi Daniel, Afrane A. N. Dankwa, Adagbe S. Philip
Supervisor:Stephen K. Armoo
Abstract:
Traditionally, the detection and curbing of overspeeding drivers are done by the deployment of service personnel on our roads and highways. With increasing technological advancements, the use of faster, accurate measures has become a necessity to prevent loss of life and limb. Some existing systems make use of handheld radar guns for speed detection. The drawback of this system is that the human factor is indispensable in this method. This system requires personnel patrolling the streets to monitor the speed of vehicles which has led to conflicts between service personnel and drivers. This project focuses on developing and implementing an innovative approach to speed detection and image processing utilizing laser pulses in order to reduce the problems of policing over speeding vehicles. Optical Character Recognition (OCR) is the process of using an optical scanner and specialized software to identify and digitally encode written or printed text. In an increasingly digitalized world, OCR allows one to decode documents. Lidar (Light Detection and Ranging) is a remote sensing technology that measures ranges by measuring the time it takes for the reflected light to return to the receiver using light in the form of a pulsed laser.
OCR and Lidar technology has been integrated into this work to assist the Motor Traffic and Transport Department (MTTD) in tracking the number plates of offending cars attempting to avoid penalties. The key components of the device are a camera for video and image capture of vehicle license plates, a Lidar sensor for distance detection, and an Arduino Uno R3 with AT Mega 3383 microcontroller for speed calculation. As a result, the fully automated speed detection device with OCR integration and deployment would limit and or eliminate the need for human intervention in vehicle speed regulation. In this report, the OCR model segmented the multiple characters of the acquired picture (number plate) and the model was trained to recognize this information when the camera captured the video of the moving car. The registration number on the plate is then routed to a database of registered cars, where the vehicle owner's information is collected. The system was tested and found to be fully functional. This project details the tests carried out as well as complications and recommendations for future work.
Keyword - OCR (Optical Character Recognition), Lidar (Light Detection and Ranging), MTTD (Motor Traffic and Transport Department).