Classification of eye movement type
|Ментор||W. Joseph MacInnes|
|Учебный семестр||Осень 2018|
|Учебный курс||2-й курс|
|Максимальное количество студентов, выбравших проект: 2-3|
What is this project?
Classification of eye movement type. Eye trackers collect data on the gaze position and pupil size at very high temporal and spatial resolution. From this data, we often want to know what state the visual system was in at any given time. The possible categories are Saccade, Fixation, smooth pursuit and Blink; and these are determined by a combination of velocity and pupil information. Choice of algorithm will likely require a solution that is sensitive to patterns over time. Open source data sets allow for comparison of classifier accuracy (Eye movement prediction and variability on natural video data sets, Dorr et al, 2012). This project for 2 or 3 students will research, implement and compare multiple classifiers to find the best performance in this task.
What will student learn?
Techniques in oculography, basic functions of eye movements, one or more classifier algorithms.
What are the initial requirements?
Basic programming skills (language is not critical, but R, C++, Java, Matlab or Python preferred)
What technologies will be used?
Open source datasets, programming.
Themes of introductory lesson?
Eye movement properties.
Directions of future development
Potential exists for publishing results at conference or journal.
Criteria for evaluation
- 50% written (English) summary of project and literature review.
- 50% code implementation and results
W. Joseph MacInnes email@example.com