The Potential of Deep Learning based Computer Vision in Warehousing Logistics

Abstract

This work describes three deep learning based computer vision approaches, that hold the potential to increase the degree of automation and the productivity of common warehousing procedures. These approaches will focus on: the re-identification of logistical entities, especially when entering and leaving the warehouse; the multi-view pose estimation of logistical entities to track and to localize them on the shop floor; and the category-agnostic segmentation of items in a bin for robotic grasping.