In order for sensor network deployments to be successful, users need to be able to derive relevant and interesting information from sensor data. Information rich multimedia data can provide high utility to users, but their interpretation is hard in resource constrained domains. We note that humans can derive information from images efficiently and propose a multi-tier camera sensor network that combines simple in-network data interpretation with a lightweight user recommendation system. We explore ways of how to decrease demands on an individual user, to balance the benefits gained from our system with the required work that needs to be put in. One of the advantages of this approach is that its cost-benefit ratio will improve as the system grows in scale, enabling techniques from large social networks, such as data recommendation based on user similarities.