3DPlanNet: Generating 3D Models from 2D Floor Plan Images Using Ensemble Methods

Research on converting 2D raster drawings into 3D vector data has a long history in the field of pattern recognition. Prior to achievement machine learning, existing studies were based heuristics and rules. In recent years, there have been several employing deep but great effort was required secure large amount for learning. this study, overcome these limitations, we used 3DPlanNet Ensemble methods incorporating rule-based heuristic learn with only small (30 floor plan images). Experimentally, method produced wall accuracy more than 95% an object similar that previous study using learning data. addition, without dimension information converted ground truth sizes 97% or more, structural form models which layers divided each object, such as walls, doors, windows, rooms, created. Using proposed generated 110,000 from end end.