A Weakly-Supervised Semantic Segmentation Approach Based on the Centroid Loss: Application to Quality Control and Inspection | Publicación