Video shooting and sharing at tourist destinations have become common activities among travelers. These videos, often enhanced with captions and audio, serve as a means to convey impressions and information about the visited locations, and they are widely disseminated through social media platforms. Such content not only enriches the personal travel experience but also contributes to promoting regional tourism and stimulating inbound demand. However, producing informative video content requires editing skills and the reliability of the information, which poses a challenge.

This study developed a new context-aware framework for caption generation in tourist video blogs by detecting tourists' points of interest (POI) using sensor fusion of GPS and camera orientation. The framework was evaluated and validated.

The results showed that by integrating camera orientation information along with geofencing, it was possible to estimate tourists' attention within the geofence with a certain level of accuracy, indicating new possibilities for guide systems.