Visual-Semantic Alignments for Automated Interpretation of 3D Imagery Data of High-Pier Bridges