Sensor network research faces challenges in exploiting vision, the most powerful of human senses. In our earlier research we presented Cyclops, a new CMOS based low power and low complexity imaging sensor with on-board computation. In this poster, we present recent advances in exploiting Cyclops in a vision network. This includes our approach to building reconfigurable and customized computations. We illustrate our systems ability to be incrementally updated after deployment, including the addition and removal of image processing libraries. This enables flexible and dynamic image processing based on the variation in requirements of the applications or in response to the environmental variations. In addition, we illustrate our progress in exploiting Cyclops in application domains. We present a tiered system which uses a stereo pair of Cyclops to locate objects in three dimensional spaces for further investigation by an actuated vision node. Finally, we demonstrate our recent deployment of a network of Cyclops for monitoring wildlife and habitat at James San Jacinto Mountains Reserve.