Towards the detection of learner's uncertainty through face

摘要

This research aims to detect uncertainty based on facial expression in a learning context with the use of the Facial Action Coding System (FACS). Although FACS has been used to categorize facial uncertainty, very few studies have worked in this field. Most studies rather focus on uncertainty detection using voice, even though uncertainty is more apparent through facial cues compared to vocal cues, and this warrants for the collection and analysis of a facial corpus. Hence, an effort to collect a facial corpus of uncertainty were made and the corpus is then analyzed. Data was collected through an experiment that entailed using stimuli to induce the uncertainty of the subject. The data was annotated in order to verify the images before proceeding to preprocessing and feature extraction techniques. The feature extraction of the images was carried out using Gabor Wavelets and classification to train the data is used Support Vector Machine (SVM).