LAUSR

Steel manufacturing is a fundamental industry in the world, and slab is one of the major steel products. In manufacturing process, individual slabs have similar shapes with different amounts of alloying elements according to production purpose or final product. Therefore, identification of individual slabs is required to prevent an incorrect production process. A widely used method in the steel industry is to inscribe a slab identification number (SIN) by using a paint marking machine, and automatic recognition of SINs is a key technology for factory automation. The objective of this paper is to develop a computer vision algorithm for recognizing SINs in factory scenes. Our automatic recognition system for SINs contains an image acquisition device that was installed at a slab yard in an actual steelworks. A factory scene collected at the slab yard contains unknown number of slabs with different sizes of SINs, and a SIN consists of 9 characters that are horizontally placed with similar distances. The first charRecognition of Slab Identification Numbers using a Fully Convolutional Network