Through-wall sensing systems can aid in performing building security, and collecting analytics or more ominously be leveraged for surveillance. With the pervasive nature of WiFi routers and devices in our office buildings and homes, we essentially place an unencrypted (at the frame level) transmitting source directly in our buildings which can then be leveraged for surveillance by adversaries. In this work, we study such a device-free WiFi sensing system for occupancy monitoring and crowdcounting and evaluate it in a number of through-wall conditions. We demonstrate that with a proper analysis of Channel State Information (CSI) collected from the WiFi signals, we can recognize both the presence of targets as well as their moving direction in a hallway environment which can be leveraged to track and count the flow of traffic throughout a building. We specifically demonstrate through real world experiments how an adversary with very limited physical access to a building can still successfully collect surveillance data of a target area through the wall.