Recent advancements in digital reconstruction have significantly improved railway bridge modeling through photogrammetry and LiDAR technologies. This study presents an integrated approach that combines UAV-based photogrammetry and LiDAR data to generate precise reality models. The application of machine learning techniques optimizes image segmentation, allowing background removal, and reducing computational complexity while enhancing accuracy. A case study involving a Portuguese railway bridge demonstrates the effectiveness of this hybrid methodology in generating photorealistic, geometrically accurate reconstructions.