<p>Product evaluation throughout the design process is a fundamental task to ensure product success. Virtual prototyping is displacing physical prototyping for product evaluation due to its lower cost and flexibility to easily generate design alternatives (colors, textures, shapes). The thesis provides a deeper understanding of the influence of the presentation medium on product evaluation. The semantic differential technique was applied in to obtain the consumers’ subjective impression when they observed furniture scenes under two different presentation mediums. High-quality realistic renderings were displayed on a computer screen equipped with an eye-tracker. The same scenes were observed by the same users (repeated measures experimental design) with a virtual reality headset equipped with an integrated eye-tracker (HP Reverb G2 Omnicept). Equivalent areas/volumes of interest were defined to calculate the eye- tracking metric dwell time. Statistical analyses then compared dwell times and values of semantic scales in the 2D and VR conditions to determine if the medium of presentation influenced them.</p> <p><br></p> <p>The experimental data obtained in the thesis confirmed that both the consumer’s subjective impression measured through bipolar pairs and the level of confidence in its assessment was influenced by the visual medium. However, the level of confidence in the assessment of a semantic scale of a product presented on VR was not affected by the sense of presence.</p> <p><br></p> <p>The amount of time (dwell time) that subjects spend looking at a specific product on a joint or individual visualization were influenced by the visual medium.</p>

Introduction: The introduction presents concepts around hands-free interactions. Furthermore, topics of digitalization, value co-creation, and how technology suppliers and end-users co-create value through the application of eye-tracking is described. Problem discussion: Healthcare is a complex system and is becoming more accustomed to the value co-creation concept with all types of stakeholders. New technologies are needed in healthcare to ensure positive patient outcomes and sterility. These technologies appear in hands-free devices such as eye-tracking technology. Limited research is found on interactions between healthcare practitioners and/or researchers with technology providers with key actors as suppliers and practitioners. Looking further at value co-creation, to achieve hands-free healthcare, it is necessary to fully utilize nascent digital technologies while incorporating them into digitalized processes. Hence, additional study is needed to investigate how key actors co-create value and promote the full use of advanced technologies. Purpose and Research Question: The purpose of this study is to understand how value is co-created by the application of hands-free devices in healthcare settings. To do that, we explore the activities performed by technology suppliers and technology end-users (healthcare practitioners and researchers) that enable value co-creation through the application of eye-tracking devices in hands-free healthcare. This study seeks to answer the research question: How do technology suppliers and end-users co-create value through the application of eye-tracking in hands-free healthcare? Theoretical Framework: Theoretical Framework was established based on scientific literature. Furthermore, it is split between concepts of value-in-use, value co-production, and two stages of digitalization, where the first stage is digitalization of products and services, the second stage is digitalization of activities and decisions. Methodology: In this thesis, qualitative descriptive research with a deductive approach is followed. Empirical data was collected through three exploratory and ten semi-structured interviews, where six semi-structured interviews were conducted with suppliers (primary data) who are employed in an eye-tracking supplying company, and four end-users (supportive data), which are healthcare practitioners and/or researchers. Findings &amp; Analysis: Here, findings gathered from primary (technology suppliers), supportive (end-users), and secondary sources (documents) were analyzed and compared to the literature and theoretical framework. Conclusion: Concluding, 20 activities were found for the value co-production part of the research and 23 activities for the value-in-use part. Some of the found activities could not be supported by scientific literature or framework and are explained as additional findings.

No description available.

Vision is an important tool for us humans and significant effort has been put into creating solutions that let us measure how we use it. Most common among the techniques to measure gaze direction is to use specialised hardware such as infrared eye trackers. Recently, several Convolutional Neural Network (CNN) based architectures have been suggested yielding impressive results on single Red Green Blue (RGB) images. However, limited research has been done around whether using several sequential images can lead to improved tracking performance. Expanding this research to include low frequency and low quality RGB images can further open up the possibility to improve tracking performance for models using off-the-shelf hardware such as web cameras or smart phone cameras. GazeCapture is a well known dataset used for training RGB based CNN models but it lacks sequences of images and natural eye movements. In this thesis, a geometric gaze estimation model is introduced and synthetic data is generated using Unity to create sequences of images with both RGB input data as well as ground Point of Gaze (POG). To make these images more natural appearing domain adaptation is done using a CycleGAN. The data is then used to train several different models to evaluate whether temporal information can increase accuracy. Even though the improvement when using a Gated Recurrent Unit (GRU) based temporal model is limited over simple sequence averaging, the network achieves smoother tracking than a single image model while still offering faster updates over a saccade (eye movement) compared to averaging. This indicates that temporal features could improve accuracy. There are several promising future areas of related research that could further improve performance such as using real sequential data or further improving the domain adaptation of synthetic data. / Synen är ett viktigt sinne för oss människor och avsevärd energi har lagts ner på att skapa lösningar som låter oss mäta hur vi använder den. Det vanligaste sättet att göra detta idag är att använda specialiserad hårdvara baserad på infrarött ljus för ögonspårning. På senare tid har maskininlärning och modeller baserade på CNN uppnått imponerande resultat för enskilda RGB-bilder men endast begränsad forskning har gjorts kring huruvida användandet av en sekvens av högupplösta bilder kan öka prestandan för dessa modeller ytterligare. Genom att uttöka denna till bildserier med lägre frekvens och kvalitet kan det finnas möjligheter att förbättra prestandan för sekventiella modeller som kan använda data från standard-hårdvara såsom en webbkamera eller kameran i en vanlig telefon. GazeCapture är ett välkänt dataset som kan användas för att träna RGB-baserade CNN-modeller för enskilda bilder. Dock innehåller det inte bildsekvenser eller bilder som fångar naturliga ögonrörelser. För att hantera detta tränades de sekventiella modellerna i denna uppsats med data som skapats från 3D-modeller i Unity. För att den syntetiska datan skulle vara jämförbar med riktiga bilder anpassades den med hjälp av ett CycleGAN. Även om förbättringen som uppnåddes med sekventiella GRU-baserade modeller var begränsad jämfört med en modell som använde medelvärdet för sekvensen så uppnådde den tränade sekventiella modellen jämnare spårning jämfört med enbildsmodeller samtidigt som den uppdateras snabbare vid en sackad (ögonrörelse) än medelvärdesmodellen. Detta indikerar att den tidsmässiga information kan förbättra ögonspårning även för lågfrekventa bildserier med lägre kvalitet. Det finns ett antal intressanta områden att fortsätta undersöka för att ytterligare öka prestandan i liknande system som till exempel användandet av större mängder riktig sekventiell data eller en förbättrad domänanpassning av syntetisk data.

Работа посвящена разработке ИТ-проекта, направленного на реализацию внедрения рекламы в игровой процесс. Для реализации функционала демонстрации рекламы, а также анализа ее эффективности, в работе исследуются методы анализа эффективности цифровой рекламы с применением технологии окулографии, методы отображения рекламы поверх игрового окна, а также методы регистрации движения глаз. Для подтверждения целесообразности внедрения описанного проекта в работе приведены полная модель предприятия, на которое планируется внедрять систему, и на основе этих данных приведен анализ экономической эффективности системы. / The work is devoted to the development of an IT project, aimed at the implementation of the introduction of advertising in the game process. To implement the functionality of advertising demonstration, as well as to analyze its effectiveness, the work investigates the methods of digital advertising effectiveness analysis using oculography technology, methods of advertising display over the game window, as well as methods of eye movement registration. To confirm the feasibility of the described project, the paper presents a complete model of the enterprise, which is planned to implement the system, and based on these data is an analysis of the cost-effectiveness of the system.

La fatigue mentale est un état complexe qui résulte d'une activité cognitive prolongée. Les symptômes de la fatigue mentale inclus des changements d'humeur, de motivation et une détérioration temporaire de diverses fonctions cognitives. Plusieurs recherches approfondies ont été menées pour développer des méthodes de reconnaissance des signes physiologiques et psychophysiologiques de la fatigue mentale. Les signes psychophysiologiques concernent principalement signaux d'activité cérébrale et leur relation avec la psychologie et la cognition. Celles-ci ont permise le développement de nombreux modèles basés sur l'IA pour classer différents niveaux de fatigue, en utilisant des données extraites d'un appareil eye-tracking, d'un électroencéphalogramme (EEG) pour mesurer l’activité cérébrale ou d'un électrocardiogramme (ECG) pour mesurer l’activité cérébrale. Dans cette mémoire, nous présentons le protocole expérimental et développé par mes directeurs de recherche et moi-même, qui vise à la fois à générer et mesurer la fatigue mentale, et à proposer des stratégies efficaces de récupération via des séances de réalité virtuelle couplées à des dispositifs EEG et eye tracking. Réussir à générer de la fatigue mentale est nécessaire pour générer un ensemble de données suivant l’évolution de la fatigue et de la récupération au cours de l’expérience, et sera également utilisé pour classer différents niveaux de fatigue à l’aide de l’apprentissage automatique. Cette mémoire fournit d'abord un état de l'art complet des facteurs prédictifs de la fatigue mentale, des méthodes de mesure et des stratégies de récupération. Ensuite, l'article présente un protocole expérimental résultant de l'état de l'art pour (1) générer et mesurer la fatigue mentale et (2) évaluer l'efficacité de la thérapie virtuelle pour la récupération de la fatigue, (3) entrainer un algorithme d'apprentissage automatique sur les données EEG pour classer 3 niveaux de fatigue différents en utilisant un environnement simulé de réalité virtuelle (VR). La thérapie virtuelle est une technique favorisant la relaxation dans un environnement simulé virtuel et interactif qui vise à réduire le stress. Dans notre travail, nous avons réussi à générer de la fatigue mentale en accomplissant des tâches cognitives dans un environnement virtuel. Les participants ont montré une diminution significative du diamètre de la pupille et du score thêta/alpha au cours des différentes tâches cognitives. Le score alpha/thêta est un indice EEG qui suit les fluctuations de la charge cognitiveet de la fatigue mentale. Divers algorithmes d'apprentissage automatique ont été formés et testés sur des segments de données EEG afin de sélectionner le modèle qui s'ajuste le mieux à ces données en ce qui concerne la métrique d'évaluation "précision équilibrée" et "f1". Parmi les 8 différents classificateurs, le SVM RBF a montré les meilleures performances avec une précision équilibrée de 95 % et une valeur de mesure f de 0,82. La précision équilibrée fournit une mesure précise de la performance dans le cas de jeu de données déséquilibrées, en tenant compte de la sensibilité et de la spécificité, et le f-score est une mesure d'évaluation qui combine les scores de précision et de rappel. Finalement, nos résultats montrent que le temps alloué à la thérapie virtuelle n'a pas amélioré le diamètre pupillaire en période post-relaxation. D'autres recherches sur l'impact de la thérapie devraient consacrer un temps plus proche du temps de récupération standard de 60 min. / Mental fatigue is a complex state that results from prolonged cognitive activity. Symptoms of mental fatigue can include change in mood, motivation, and temporary deterioration of various cognitive functions involved in goal-directed behavior. Extensive research has been done to develop methods for recognizing physiological and psychophysiological signs of mental fatigue. Psychophysiological signs are mostly concern with patterns of brain activity and their relation to psychology and cognition. This has allowed the development of many AI-based models to classify different levels of fatigue, using data extracted from eye-tracking devices, electroencephalogram (EEG) measuring brain activity, or electrocardiogram (ECG) measuring cardiac activity. In this thesis, we present the experimental protocol developed by my research directors and I, which aims to both generate/measure mental fatigue and provide effective strategies for recuperation via VR sessions paired with EEG and eye-tracking devices. Successfully generating mental fatigue is crucial to generate a time-series dataset tracking the evolution of fatigue and recuperation during the experiment and will also be used to classify different levels of fatigue using machine learning. This thesis first provides a state-of-the-art of mental fatigue predictive factors, measurement methods, and recuperation strategies. The goal of this protocol is to (1) generate and measure mental fatigue, (2) evaluate the effectiveness of virtual therapy for fatigue recuperation, using a virtual reality (VR) simulated environment and (3) train a machine learning algorithm on EEG data to classify 3 different levels of fatigue. Virtual therapy is relaxation promoting technique in a virtual and interactive simulated environment which aims to reduce stress. In our work, we successfully generated mental fatigue through completion of cognitive tasks in a virtual simulated environment. Participants showed significant decline in pupil diameter and theta/alpha score during the various cognitive tasks. The alpha/theta score is an EEG index tracking fluctuations in cognitive load and mental fatigue. Various machine learning algorithm candidates were trained and tested on EEG data segments in order to select the classifier that best fits EEG data with respect to evaluation metric ‘balanced accuracy’ and 'f1-measures'. Among the 8 different classifier candidates, RBF SVM showed the best performance with 95% balanced accuracy 0.82 f-score value and on the validation set, and 92% accuracy and 0.90 f-score on test set. Balanced accuracy provides an accurate measure of performance in the case of imbalanced data, considering sensitivity and specificity and f-score is an evaluation metric which combines precision and recall scores. Finally, our results show that the time allocated for virtual therapy did not improve pupil diameter in the post-relaxation period. Further research on the impact of relaxation therapy should allocate time closer to the standard recovery time of 60 min.

In this thesis the Tobii eye-tracker 4L was used to investigate how well eye-tracking solutions such as a confirmation-click and dwell-time algorithm compares to the standard mouse input device when performing selection tasks in a map environment. In order to distinguish the different complexity one could face, two user cases are proposed. Scenario one includes non clustered objects. Scenario two include clustered occluded objects. A user study with nine different participants where conducted in order to compare the execution times and find out how error prone the different methods were. Each test participant performed eight different tests, three in the non-clustered scenario and five in the clustered scenario. In two of the tests in the clustered scenario test participants were aided with zooming through a zoom algorithm. The methods was evaluated by calculating the average execution times and errors along with the corresponding standard deviations. In order to grasp the users experience a subjective cognitive load score was calculated with the help of a questionnaire. The eye-tracker methods was found to be competitive in comparison to mouse interaction in the more simple non-clustered case. However, in a more complex scenario such as the clustered case the mouse interaction had the lowest average completion time and cognitive load score. A different type of selection behaviour was discovered among the test participants in the clustered scenario due to the difference in precision between the eye-tracker and mouse interaction. Finally interesting areas to consider in the future is presented and discussed. / I denna avhandling användes en Tobii eye-tracker 4L för att undersöka hur väl eye-tracking metoder så som en bekräftelseklick och dwell-time algoritm jämför sig med standard mus interaktion vid objekt selektion i en kartmiljö. För att urskilja variationen i komplexitet man kan möta föreslås två olika användarfall. Scenario ett inkluderar objekt som är distinktivt separerade och därav ej grupperade. Scenario två inkluderar grupperade samt ockluderade objekt. En användarstudie med nio olika deltagare genomfördes för att jämföra exekveringstiderna och ta reda på hur felbenägna de olika metoderna var. Varje testdeltagare utförde åtta olika tester, tre i det icke-grupperade scenariot och fem i det grupperade scenariot. I två av testerna i det grupperade scenariot fick deltagarna hjälp med att zooma genom en zoomalgoritm. Metoderna utvärderades genom att beräkna de genomsnittliga exekveringstiderna samt antal fel tillsammans med motsvarande standardavvikelser. För att förstå hur användarna upplevde de olika metoderna togs en subjektiv kognitiv belastningspoäng fram genom ett frågeformulär. Eye-tracker metoderna var konkurrenskraftiga i jämförelse med musinteraktion i det enklare fallet där objekt ej var grupperade. I ett mer komplext scenario, såsom i det grupperade fallet, hade dock musinteraktionen den lägsta genomsnittliga exekveringstiden och kognitiva belastningspoängen. En annan typ av selektions beteende upptäcktes bland testdeltagarna i det grupperade scenariot på grund av skillnaden i precision mellan eye-trackern och musinteraktionen. Slutligen presenteras och diskuteras intressanta områden att överväga vid framtida arbeten.

The growing numbers of people suffering from Alzheimer’s and other dementia related diseases are expected to accelerate, and the cost for these diseases in Swedish healthcare is high. There are many ongoing research projects in the dementia diagnostics field which aim to detect cognitive impairment at an earlier stage, which would result in reduced costs in healthcare and improved life quality for sufferers. This work aims to investigate if it is possible to classify cognitive impairment based on a person’s eye movements. More specifically, it will explore the possibility of automating an established picture description task that is widely used in traditional dementia diagnostics. In order to do this, eye tracking data was collected during numerous conductions of this task. The eye tracking data was then parsed in to eye movement features and Binary Logistic Regression was used to classify these eye movements. The results showed that the average accuracy of the classification reached 73%. The results did not confirm that eye tracking technique can be used to automate neuropsychological test with an accuracy high enough, but to use a machine learning approach for detecting deviances in eye movement patterns appears to be a promising approach. Furthermore, this work analyzes the possibilities for practically implementing eye tracking techniques in Swedish healthcare in order to detect cognitive impairment at an earlier stage. Provided that an eye tracker can detect cognitive impairment with an accuracy equal to or higher than a medical professional can maintain, the study argues that automated neuropsychological tests at health clinics could be the key to detect cognitive impairment at an earlier stage. / Antalet personer som lider av alzeimers och andra demensrelaterade sjukdomar förväntar att öka med accelerande fart och kostnaden for dessa sjukdomar för svensk sjukvård är hög. Det finns idag många pågående forskningsprojekt inom demensdiagnostik där man analyserar personers ögonrörelser för att kunna detektera kognitiva svårighetet i tidigt stadie. Forskningen görs för att minska på kostnaden och öka livskvaliten för de som insjuknar. Detta arbete syftar till att undersöka om det går att använda maskininlärning för att klassificera kognitiv svårighet baserat på en persons ögonrörelser. Mer konkret vill man undersöka om en automatisering av en etablerad bildbeskrivningsuppgift, som idag används flitigt inom demensdiagnotistik. Det har äarför samlats in data som representerar personers ögonrörelser under tider de utför olika demenstester. Med hjälp av datan har man sedan tagit fram olika synfunktioner och använt binär och logistisk regression för att klassifisera dessa ögonrörelser. Det genomsnittliga resultatet visade att modellen klassificerade rätt i 73% av fallen. Detta resultat kan inte bekräfta att denna teknik kan användas för att automatisera neuropsykologiskt tester med tillräckligt hög noggrannhet. Daremot ser det lovande ut att kunna ända maskininlärning för att detektera avvikelser i ögonrörelser, hos personer som lider av kognitiva svårigheter. Vidare analyseras också möjligheten att praktiskt implementera tekniken där man analysera ögonrörelser i svensk sjukvård. Resultatet visar att om det är m jligt att utforma en modell som diagnotiserar bättre än vad en professionell läkare gör, så går det att argumentera for att automatiska, neuropsykiska tester skulle kunna vara en nyckel för att detektera kognitiva svårigheter i tidigt stadie.

No description available.

No description available.