<![CDATA[Event satisfaction is commonly measured through post-event surveys, yet such instruments are retrospective and vulnerable to memory bias. This study examines whether event documentation photos can be used to construct an auditable satisfaction proxy through computer vision. The workflow combined media inventory, duplicate removal, face quality filtering, face re-detection, face cropping, and smile-based expression proxy scoring. From 1,835 documentation files, 1,739 JPG photos were identified as the main image source. After quality screening, 236 photos were categorized as suitable or moderately suitable for expression analysis. Among them, 178 photos (75.42%) were classified as positive smiling, 56 photos (23.73%) as neutral or non-smiling, and 2 photos failed face re-detection. The average satisfaction proxy score reached 0.653 on a 0-1 scale, indicating that the atmosphere of Howell Expo 2025 was generally positive. The findings suggest that the event was sufficiently successful in creating a positive visitor experience, although the resulting score should be interpreted as an operational proxy rather than a direct measurement of individual emotion.]]>
Copyrights © 2026
