| dc.description.abstract | Many of today's CBT type examination systems use input devices, such as keyboard
and mouse, for inputting or choosing answer. This has been the concern as this practice is
not safe -- by noting that examinee has full control of input devices on examination site that
enables making fraude.g. trying to access answer key folder or by manipulating the
examination system to reaccess the examination system the second time, and increasing
the risk of broken devices .e.g. unintentionally or intentionally dropping mouse or other
input devices that will affect the examination process.
Thus, writer has developed a touchless examination system creation which uses only a monitor layer and web camera, by implementing Computer Vision to enable hand gesture detection and recognition as interactive input with quite high accuracy and quite low latency.
This system is developed by implementing Python's Computer Vision libraries, which use Convolutional Neural Network to extract, detect, and identify hand landmarks so that computer can know the feature being accessed by examinee on the monitor layer.
As the result, writer has successfully developed a prototype of touchless CBT system which supports examination in Multiple Choice or Essay form, timing and additional picture display, percentage of question completence informing, and self scoring mechanism with self report mechanism. Furthermore, the evaluation results of system's accuracy and latency reveals that this system is proper to be real-implemented. | en_US |
