Automatic Classification Methods of Photographs Based on Geographical Attentions of Crowd

We propose a digital photographs automatic classification method by its subject that has attracted crowd´s attention. Our method can classify digital photographs without using pixel information. When an interesting event occurs, many people take photographs of the event. If we could collect digital photographs of the event and build an image database quickly, then we are able to share information in real-time with high accuracy. However, a simple method that transmits the image data may causes congestion of communication network. In this paper, we propose a highly available classification system. Our method estimates subjects of digital photographs using only information about a state of a camera when photographed, and classify the photographs based on the subject. We can suppress congestion because the system preferentially transmits photographs that meets a browse request of viewers. We verified an effectiveness of our method by an experiment using a commercially available digital camera.