Data-Driven Interior Plan Generation for Residential Buildings in Vietnam

We propose a new data-driven technique to automatically and efficiently generate floor plans for residential buildings in Vietnam with certain boundaries. We focus on improving the accuracy of the algorithm that was developed in another country to match the architectural style in Vietnam. Along with collecting a large number of architectural plans in Vietnam, we have proposed a method of semi-automatic data labeling using CNN deep learning network. As the result, we achieved the overall accuracy over 85% for prediction room type and room location on the test dataset.