AI-mediated Job Status Tracking in AR as a No-Code service

Andrey Konin, Shakeeb Siddiqui, Hasan Gilani, Muhammad Mudassir, M. Hassan Ahmed, Taban Shaukat, Muhammad Naufil, Awais Ahmed, Quoc-Huy Tran, M. Zeeshan Zia

Abstract: Sophisticated AR experiences that can track the status of a manual assembly or maintenance job and provide corresponding guidance are built by teams of computer vision engineers over several months. These engineers model temporal causation on top of object detection and human skeletal tracking provided by traditional AR and AI SDKs. We demonstrate our Pathfinder system which automatically builds computational models of a complex manual task, such as a manufacturing assembly activity, given recorded demonstrations of the task - without requiring any custom coding. These computational models enable ordinary AR developers to build AI-mediated feedback in minutes instead of months.