Measures to improve outdoor crowdsourcing photo collection on smart phones

Visual crowdsensing (VCS) uses built-in cameras and sensors for smart devices to take photos of interesting objects or views. A large number of VCS tasks are location-constrained outdoor tasks, such as taking photos of a landmark, and their rewards are usually higher than indoor tasks since they are very labor-intensive and time-consuming. To reduce the time consuming of accomplishing crowdsourcing photo collection tasks, we focus on optimizing the task allocation method and improving the photo-taking client. This paper introduces our VCS platform for collecting outdoor crowdsourcing photos, namely LuckyPhoto. Firstly, we use different task allocation methods to handle different types of tasks. For instance, we propose a novel task allocation method for allocating multi-facet tasks that require photos taken from an object's peripheries. Secondly, we develop Apps for both data requesters and workers. Both Apps utilize sensors in smartphones, which brings convenience to data requests to create tasks and to workers to accomplish tasks respectively. As a result, LuckyPhoto App provides workers with hints to collect photos efficiently. 78 college students were paid to use LuckyPhoto to accomplish tasks and 749 photos were collected. Experimental results show that i) the task allocation ratio was increased and ii) workers can accomplish VCS tasks more efficiently.