Application of optical character recognition with Tesseract in logistics management

Abstract

Warehouse and inventory management poses many challenges, for example in carton handling when upstream suppliers use labelling systems that are incompatible with a company's downstream system. In such cases, information is digitised using manual labour: this process can often become a bottleneck and, eventually, a source of handling errors. In this paper, the feasibility of applying optical character recognition (OCR) technology in carton handling is assessed, and a prototype based on the open-source engine Tesseract is described in detail. Its performance on both printed and handwritten text is quantified, as well as the impact of turning the problem into a matching problem rather than a pure recognition problem.