<![CDATA[The growth of optical e-commerce has created a need for interactive tools that let consumers virtually try on eyewear before purchasing. VTO may be a solution, but its value depends on face-detection performance and user satisfaction. In this study, we develop and evaluate a web-based VTO system using the Haar Cascade Classifier, focusing on technical performance under different facial conditions and perceived user experience. We acquired the images with a webcam and converted them to grayscale. We detected faces, located the eye regions, and overlaid the eyewear in real time. The approach was experimental-developmental. We assessed technical performance across 15 predefined trials varying lighting, face orientation, occlusion, and number of faces. User experience was measured by the 26-item User Experience Questionnaire (UEQ). The system was successful 11 times out of 15 trials. That is a success rate of 73.33%. Most failures were found in non-frontal facial orientations and low lighting conditions. The UEQ results showed positive user perception with average scores of 2.18 for Dependability, 2.14 for Attractiveness, 2.04 for Pragmatic Quality, and 1.99 for Hedonic Quality. Perspicuity was the lowest-scoring (1.93), which might be an area to improve for clarity of the interface. The results show that the VTO based on Haar Cascade is feasible under the tested conditions, but the detection robustness has its limitations. Further research should compare alternative detection and facial landmark methods, evaluate eyewear alignment accuracy, and examine performance across more diverse conditions and users.]]>
Copyrights © 2026
