| Abstract [eng] |
In this master thesis, we aimed to investigate humans’ fatigue detection by using keystroke dynamics. The aim of this research was to analyze existing approaches of humans’ fatigue detection, suggest a more human friendly detection method, implement and test it. Furthermore, we analyzed these systems’ results and drew appropriate conclusions. The research consists of three main parts: analysis, designing and implementing testing systems. In the analysis part, we analyzed how fatigue can affect humans’ behavior, how it can be detected and also how current detection methods are implemented. Based on the background of our analysis, we hoped to suggest a more human friendly fatigue detection method. In the designing part, we designed an application which can detect keystroke dynamics, store it in database and determinate humans’ fatigue. We implemented this application using Microsoft Visual Studio 2015 and C# programming language. To store our data, we used Microsoft SQL server 2016 database. In the experimental part, the research of the user’s keystroke dynamics data classification is being carried out. Carrying out the experiment, we collected four participants text typing data for two weeks. With the collected dataset we evaluated to what precision we could identify human’s fatigue. According to the results of our approach, we can achieve an accuracy of 89,79 % of identifying a human’s fatigue. |
