Abstract
An embedded vision architecture for the detection, identification, tracking and pose estimation in the 3D space of different types of pallets in warehouse management and logistics is presented here. The proposed hardware and software architecture was implemented in the context of the EV-Lib funded project and is able to provide a fully working system for monitoring and assessing the types, 3D positions and locations of pallets and boxes in real-time in a warehouse environment or distribution center, based on a vision system infrastructure and edge artificial intelligence methodology.