Supervision |

Designing and Implementing Voyager - An Intelligent Travel Companion

Fri, 26 May 2023 00:00:00 +0000

Abstract

This paper explores the development of a user-friendly mobile application that combines travel planning and itinerary management. By comparing existing apps on the market and conducting user research, we identify user needs and ways to differentiate our app from competitors. Our app incorporates state-of-the-art AI technology to generate personalized itineraries for users visiting Copenhagen. We present the results of user involvement, including user interviews and usability tests, and analyze their feedback. In addition, we compare similar apps on the market and discuss design choices for our app’s user interface. We provide detailed information on the technical implementation of our app and explore future possibilities for development and integration with AI. Overall, this paper provides insights into the creation of a travel planning and itinerary management app that offers a unique and user-friendly experience for travelers. Our approach combines user research, innovative technology, and thoughtful design choices to create an app that stands out in a competitive market.

Mobile Games for the Visually Impaired

Fri, 26 May 2023 00:00:00 +0000

Abstract

Grand View Research predicts the gaming industry to grow at an annual rate of 12.9% from 2022 to 2030. However, not everyone can enjoy everything the gaming industry provides. People can have many impairments that prevent them from enjoying a video game. There are many possible guidelines for developers to follow to improve accessibility, but they lack a focus on activating other senses than sight. This project shows the development of an application that implements features that activate multiple senses. The application is then tested on 25 individuals across 5 groups. The data analysis presents partial results and tendencies because of a lacking data set. The goal is to collect a ten times larger data set and perform a Gaussian analysis before publishing the results of this project in a conference paper.

Retrieval, Visualization, and Analysis of Graffiti in Copenhagen

Fri, 26 May 2023 00:00:00 +0000

Abstract

This project investigates the relationship between the occurrence of graffiti and various factors in the Amager districts of Copenhagen. The research is focused on social factors such as population density and income groups. However, the relation to crime and areas with graffiti is also examined. Through geospatial analysis and machine learning models, the project explores patterns and correlations associated with graffiti. The geospatial analysis showed that certain areas have a higher concentration of graffiti, with urban areas exhibiting more compared to residential areas. The machine learning models showed limited success in predicting the occurrence of graffiti solely based on income and population density but achieved moderate accuracy in identifying graffiti tags. The findings suggest that factors beyond income and population density may contribute to graffiti occurrence. Further research is needed to explore additional factors and improve the predictive models. Overall, this project provides valuable insights into the distribution and potential influencing factors of graffiti, contributing to a better understanding of this urban phenomenon.

Correct Execution of Weightlifting Exercises using Pose Estimation

Wed, 11 Jan 2023 00:00:00 +0000

Abstract

This thesis is a technical analysis of the weightlifting exercise deadlift, using pose detection and machine learning. Weightlifting is an increasingly popular exercise method with substantial benefits. However, if done incorrectly could lead to injuries. The project aims to research whether machine learning technologies help create a solution that is an alternative to hiring a personal trainer. The technical goal is to recognize correct and incorrect movements from video input by running videos through Google MediaPipe pose detection to gather x, y, and z coordinates. The dataset contains either correct or incorrect video labels to feed the machine learning prediction model to predict whether or not a particular video was correct or incorrect. By doing this, the trained model can clear distinct between correct and incorrect movement, resulting in an alternative to hiring a personal trainer.

Development of App to the Restaurant Business

Wed, 24 Aug 2022 00:00:00 +0000

Abstract

This project researches whether Machine Learning (ML) and Social Media can help improve the experience for a customer who dines at restaurants. Previous studies have shown that restaurants should make their reservation and payment systems digital and online to increase customer service. Our project focus on the User Experience of the restaurant experience. We found out that ML and Social Media can contribute positively to the experience by focusing on the user-to-restaurant interaction rather than the user-to-user interaction. The limited user-to-user elements should focus on restaurant reviews and planning events to which ML can contribute positively, by identifying food images and connecting them to menu items on a menu card. This project discusses various ML and Social Media elements and how to utilize them, relying on data collected through questionnaires and interviews. Though marketing for businesses is not within this project’s scope, the test group mentioned that they would rather receive offers and news from restaurants than reading user updates, which introduces an exciting angle to how restaurants could use the proposed system further. The business aspects still need research, exploring how restaurants can use the proposed system for marketing themselves and how beneficial ML would be for business owners.

Using Eye/Gaze Tracking (With Narrator) to Improve Reading Ease, Speed, and Comprehension

Wed, 17 Aug 2022 00:00:00 +0000

Abstract

This bachelors project covers eye tracking and how it can be used as a interface with a Text-to-Speech narration, the end goal being providing assistance to those who struggle with reading this could be people with reading learning disorders such as dyslexia.

Most estimates of people with reading learning disorders are between 5-20% accounting for a substantial amount of people that struggle with reading and with how integrated reading is in our societies it makes sense to try and develop assistance for those that struggle.

The project goes into detail about how the different technologies can interface to increase interactability of narration software especially during narration.

The results being that the sample size and bias within the data results in data of to little quality to prove anything regarding the hypothesis.

This exact field of research is not directly being explored, though the different aspects are. In the end this project tries to unify and explore a lot of ideas and cannot reach any resolution.

Correct Disc Golf Form: Classification of the Backhand Throw using Neural Networks

Fri, 24 Jun 2022 00:00:00 +0000

Abstract

Form is essential when analyzing and reviewing a backhand disc golf throw. The form defines if the throw is performed correctly and the poses of the body define the form. By looking at the body poses the throw can be classified, critiqued, and improved upon. The form consists of different motions which are analyzed using 3D data collected using machine learning solutions on a data set of recorded disc golf throws. By processing the 3D data from recorded throws the form is classified into three classes that represent the start, mid, and end of the throw. The three classes are shown as clusters using Principal Component Analysis (PCA). The PCA showed more overlapping clusters for the start and middle of the throw compared to the end. Classification solutions include a variation of trained LSTM networks and a solution using MediaPipe Pose Classification. The paper concludes that LSTM models perform faster and more accurately than the solution using MediaPipe Pose Classification when analyzing disc golf throws. However, the classification only provides insight for classifying the different forms and not the quality of form.

Using Machine Learning to Improve the Whiteboard Experience

Thu, 23 Jun 2022 00:00:00 +0000

Abstract

Virtual meetings and conferences are getting more common and more mainstream in the workplace. This shift in use of technology means that other work practices has to adapt to work with virtual meetings. One such practice is the use of writing on whiteboards. Just like some people prefer to read from a book instead of a monitor, a practice like writing on a whiteboard might never be replaced by writing on a tablet. This creates a set of problems of how can the whiteboard be integrated to work with the virtual world. There’s many ideas and potential solutions for this like using text recognition, but most of these solutions require the whiteboard to be detected in the first place. To solve this issue, a whiteboard detection model is proposed which is composed of a convolutional neural net to classify whiteboards in real-time videos through semantic image segmentation and computer vision to process the outline of the classified whiteboards into a set of points which can be used for further analysis and processing.

Opportunities with Hand Gesture Technology in Mobile Gaming

Mon, 23 Aug 2021 00:00:00 +0000

Abstract

During the past decade, there has been steady developments in the area of computer-vision based hand-gesture recognition (HGR) technologies, and expansion in the environments they are available for. Hand-gesture input, combined with head-mounted displays, has become the principal interaction method in virtual reality games. It also shows promise in other areas, such as sign-language recognition, interactive museum exhibitions, and interactive displays available in public spaces. This paper explores the possible introduction of HGR-based interaction in mobile games, based on the identification of key concepts in literature examining the aforementioned areas. The result is a proposition of four general heuristics guiding the design and development of mobile games based on HGR as the primary interaction method.

Head Pose Estimation using Deep Learning

Tue, 08 Jun 2021 00:00:00 +0000

Abstract

Estimating the head pose of a person is a significant task with various applications such as helping with fitting 3D objects to the face, user interaction in computer systems or applications, and tracking the driver’s direction of view. There are two practical approaches to estimating the head pose based on an image of a human head. One is to estimate the landmarks of the human face and then solve the 2D to 3D correspondence problem with a 3D human head model, also called a landmark-based approach. Another method is to train a Deep Convolutional Neural Network to predict the head pose directly from the image of a human head, also called an appearance-based approach. An example of an appearance-based approach is the architecture called HopeNet, which was presented in “Fine-Grained Head Pose Estimation Without Keypoints” and produced state-of-the-art results when compared to commonly used landmark based approaches. However, this method was trained on synthetic datasets and only tested on one dataset with actual data (BIWI) due to the lack of other real datasets of human heads corresponding to Euler angles. The 2018 iPad Pro has a TrueDepth sensor, which, together with the Apple Neural Engine, can capture detailed information from the human face and extract the head pose from this. We present a new approach to create a custom dataset with images of heads and corresponding intrinsic Euler angles (pitch, yaw, roll) and 3D position by recording videos with a 2018 iPad Pro and collecting the frames with the corresponding Euler angles. We show that this data collection method and the HopeNet architecture show great potential for head pose estimation by producing prediction errors similar to what the 2018 iPad Pro produces.

Object Tracking System (VidIT)

Fri, 04 Jun 2021 00:00:00 +0000

Abstract

This thesis investigates, if an IT product can increase learning in an online setting. Information is included in regards of learning and the development of VidIT, which is an automated tracking system powered by a smartphone and an Arduino. The system can track people with the help of a motorized pan tilt mount. The purpose of VidIT is to enhance learning during COVID-19, by enabling students and teachers to record themselves single-handily while moving around. A survey, a user test and a performance test was conducted to gather data on the current situation of teaching in an online setting, testing of the usability and performance of VidIT. Based on the tests, it was concluded that the resulting system worked as intended. However, some improvements are needed to effectively improve learning and teaching in an online setting. These improvements includes but are not limited to, streaming functionality, movement prediction and faster computation in relation to the objection detection algorithm.

Building a Gamified Habit Application

Mon, 17 Aug 2020 00:00:00 +0000

Abstract

A large part of human behaviour is a function of habits, and thus habits have a notable influence on the well-being of the individual. Although many mobile applications claim to support habit formation, there is a lack of habit applications that are based on theory of habit formation. Gamification shows promise for supporting long term behaviour change such as habit formation but is not prevalent in habit applications. Gamification is a relatively new trend that focuses on applying game techniques to non-game contexts in order to provide motivational benefits. This paper intends to elicit the design of an application that aims to support the formation of habits. Two literature was therefore conducted on habit theory and gamification. Based on these results, an initial set of requirements and a prototype was created. This prototype was used to gather feedback from three potential users leading to nine final requirements grouped into four major categories concerning the daily tracking of habits, long term goals, implementation intentions and rewards. An MVP of the application was developed based on these requirements, the prototype and user feedback.

Machine Learning in Android Applications

Fri, 29 May 2020 00:00:00 +0000

Abstract

This project is motivated by the ever-increasing popularity of machine learning techniques for solving repetitive everyday tasks, as well as the availability and importance of smartphones in today’s society. The combination of the two creates an environment in which the use of machine learning for simplifying mundane tasks in mobile applications may be experimented with. This project is a study in the use of machine learning in the context of such a mobile application, and specifically uses the Firebase ML Kit mobile SDK in that pursuit. The project includes the development of an application that allows users to generate descriptions of electronic devices they wish to post for sale on online marketplaces. The application utilizes machine learning and natural language generation to present the user with a textual description of the image submitted to the application. This project gives an overview of central machine learning principles, and goes into detail about the concepts relevant to solving the problem in question, namely classification, and neural networks. It also describes the process of implementing the application and how Firebase ML Kit provides machine learning capabilities, as well as how SimpleNLG provides natural language generation functionality to the application. The project further reflects on the application created and the use of ML Kit therein.

Implementation of the Progressive Web App - Woodle

Thu, 28 May 2020 00:00:00 +0000

Abstract

This project aims to create a native-like application on the web to test the current capabilities of Progressive Web Apps. Woodle is a Progressive Web App, which is an application run on the web, but with enhancements to create a native-like application. Progressive Web Apps bring native-like capabilities, such as geolocation, push-notifications, offline use, and more to web-platforms. Therefore, in Woodle, users can register an account and track the activities through GPS location. After completing an activity, it’s saved so the user can keep track of activity stats and previous activities. Users can add friends and see their activity history as well.

Road Safety with Android Auto and Machine Learning

Tue, 26 May 2020 00:00:00 +0000

Abstract

This thesis aims to research the question of how to predict road safety and how a driver can safely receive relevant information on road safety during a drive. This has become a relevant field of research, with sophisticated computing hardware available as a feature in cars. Additionally, operation areas and computation capability of mobile devices are expanding. The results of the experiment in this thesis has been an Android application which implements Machine Learning Models and Statistical Models to predict accidents, based on the current situation of the user. The Machine Learning Models do not provide valid scientific evidence for the predictions to be correct, due to the supervised historical traffic data, used to train the Machine Learning models, having inconsistent patterns of how accidents happen. The Machine Learning models are activated by Statistical Models using historical traffic data. The models are only compatible to some extent. This is limited by a historical weather data set, which only enables the model to predict accidents within a range incorrect with a level of abstraction. Thus the Statistical Models and the Machine Learning Models are implemented in the application using the Android System compatible with the Android Auto subsystem. Android Auto enables a safe communication channel with the drive. The application is distributable to Android Users and compatible with 60.3% of all android devices. In the future the models predictions might be invalid, as the behaviour of a car might change. Although the experiment does not provide any sophisticated pipeline for extending the models with new data.

Highway Monitoring System

Mon, 14 Jan 2019 00:00:00 +0000

Abstract

The future of traffic calls for intelligent transport systems, which can handle the increased demands and expectations of future traffic. This thesis examines the possibilities for creating a highway monitoring system. Highway monitoring is important in the future because traffic will increase. The purpose of this thesis is to build and evaluate such a system with focus on high performance and accuracy while respecting the privacy of drivers on the highway. It presents the design of the system based on different image processing and machine learning methods, with evaluations of their individual performance and accuracy, as well as the overall performance of the system. The results show that it is possible to build a highway monitoring system on relative mundane hardware, while still attaining a specific accuracy above 90% in most cases and a generalized performance of above 100 FPS.

This thesis contributes to the ongoing research in intelligent transport systems by proposing a coherent highway monitoring system with great performance and strong privacy principles. It shows that privacy does not have to impact the functionality or performance of a highway monitoring system. It serves as an indication that privacy is a topic that is worth researching in combination with different systems. This thesis also discusses the implications of the proposed system for further investigation in validating and improving performance and in how different types of data collectors can work together to form a coherent highway monitoring system.

Automated Lecturer-Tracking System

Mon, 10 Sep 2018 00:00:00 +0000

Abstract

Development of technology has brought some significant changes to the educational system, resulting in some new means of gathering the knowledge. In this project, we focus on the video lectures that provide significant benefits for all the students. We aim to enhance the way video lessons are recorded by introducing the system for automatic lecturer-tracking. This thesis introduces the new approach in implementing an automated lecturer-tracking system by using a smartphone as the replacement for a camera device and a processing unit. The proposed solution uses the YOLO real-time object detection system and tracking algorithms from iOS Vision framework to detect and track the lecturer. A motorized pan-tilt head rotates the smartphone based on the input the smartphone sends to it. Experimental results show that the system can perform the desired behavior of lecturer-tracking, eliminating the need for human help in the process of recording.

Computer Vision til EvoBotten

Mon, 13 Jun 2016 00:00:00 +0000

Abstract

Automated production using robots play a significant role in chemistry, biotechnology and microbiology. Robots that are designed to perform a specific task are in the long run cheaper and much more efficient than humans. In research, however, most tasks are done by humans even though many tasks are cumbersome and repetitive. What complicates the introduction of robots in research are the small variations in the task sequences that are frequently introduced. In this Master thesis we will contribute to a project called the EvoBot which seeks to make robots an integral part of research in order to lower costs and speed up the process of experimenting. We develop a proof-of-concept for a computer vision application for the EvoBot that enables it to find, locate and classify petri dishes and well plates. We present the design of the system implemented using the OpenCV framework along with physical modifications to the EvoBot. In addition to the vision system we develop a framework to test its precision and accuracy and lay ground work for future improvements. We evaluate different approaches based on accuracy and precision results of the detection methods that are experimented with. The evaluation indicates that detection without the use of tagging is feasible for use in the industry, with the introduction of some future improvements.