Systems and methods are disclosed for target-based AR devices to perform low-power front-end passive scanning of targets to alert users to AR content linked to any image targets the users may be viewing. Passive scanning by AR devices relieves users of the need to manually activate a camera for AR image target identification, and helps to identify image targets the users may be unknowingly viewing. To conserver power, AR devices may autonomously activate a camera to perform an exploratory scan when the AR devices detect from users' movement patterns that users may be interested in certain targets or is in a state of attentiveness. AR devices may identify one or more image targets from the exploratory scans. If users elect to interact with the AR content, AR devices may activate the camera to perform a full capture or real-time tracking of the image targets to augment the AR content.