Relevance Metrics for Coverage Extension Using Community Collected Cell-phone Camera Imagery

摘要: The rapid increase of mobile phone cameras has enabled users to easily take and share pictures. This created a potential for device driven sensing our world at previously unachieved spatio-temporal granularity, enabling variety new applications. data collection activity is highly uncoordinated hence, key issues in effectively using such imagery understanding the relevance value each image. Having can not only streamline resource usage sharing image but also support development incentive mechanisms contribute worthwhile data. We discuss problem assigning values images from devices with respect an application’s existing data-set. describe general information theoretic framework computing relative specific computation coverage based metric. develop practical algorithm compute methods make scalable large sets. Finally, we present prototype implementation demonstrating methods  on real