CrowdPic: An Interactive and Selective Picture Collection Framework for Participatory Sensing Systems

Taking pictures is a commonly-seen participatory sensing manner, and has been explored in many application areas. However, in a participatory image collection system, people are sensitive to the cost of network transmission and power, which may impact their willingness to participate. Moreover, in crowd image sensing, redundant data with varied quality may be contributed by different participants, which should be carefully processed and selected at the server side. In this paper, we propose Crowd Pic, a generic framework for mobile picture crowd sourcing. In Crowd Pic, in order to increase participants' willingness, we leverage multi-round interactions between clients and the server to reduce the redundant data transmission. Meanwhile, we use the sensors of the mobile devices to evaluate the quality of pictures, and select the best picture to be used. Experiments over a dataset collected by 38 recruited participants validate the effectiveness of our framework.