1. Introduction
In the steel industry, recognizing BIN is used for the categorization of steel products during steel product manufacturing. This process is crucial because it prevents steel products from mixing during the manufacturing process. Due to that, many algorithms were studied to predict product identification numbers. In this paper, we adopted a semantic segmentation network for BIN recognition with high reference on [1], [2], [3], [4], [5], and [6] with several tunings. This network did BIN recognition with over 99% accuracy on each factory dataset.