Confidence plays a crucial role in effective communication, as research indicates that a speaker’s assurance can be discerned through visual cues such as eye contact and head movements. These cues reflect confidence levels, which are pivotal in domains like online interviews, education, and healthcare. In our study, we analyzed videos of 65 participants, categorizing their responses as either high confidence or low confidence. By extracting gaze and head pose features, we employed several machine learning algorithms for classification. Notably, Naïve Bayes emerged as the most effective model, achieving an impressive 88% accuracy, highlighting its potential in discerning confidence levels based on visual cues.