Abstract The proper placement of visual sensors across a sensor field for covering targets with arbitrary location and orientation is a mission-critical decision in surveillance applications. The specifics of sensor deployment in these applications not only determine the maximum achievable coverage, but it also affects the quality of the target’s appearance in cameras for subsequent use in vision tasks. However, the inaccuracies inherent in localization techniques and the lack of knowledge regarding the target orientation render existing proposals insufficient for real-life scenarios. In this paper, we address both challenges. First, we extend the conventional point representation of targets with a circular model to account for full-angle coverage of targets with known location yet with unknown orientation from all directions. We then assume, in the absence of precise location information, a trajectory profile of targets could instead be generated through the importance sampling of the environment, indicating spots where the target is most likely located. This profile-based abstraction enables us to capture the uncertainty in target’s location by encircling every agglomeration of proximal samples within one cluster. Each cluster can then be viewed as a virtual macroscopic circular target for which we formulate the coverage problem in terms of a Binary Integer Programming (BIP) model. We have also taken into account the presence of obstruction in between multiple targets by calculating the angles of view of the sensors in an occlusion-dependant manner, effectively determining optimal placement for maximal instead of full-angle coverage. Evaluation results derived from our simulation experiments reveal that the proposed mechanism can effectively achieve high coverage accuracy with minimum number of deployed sensors.