Abstract: In order to generate reasonable furniture layouts in complicated indoor scenes automatically, we proposed a data-driven method for automatic indoor areas layout calculation. Given examples of sensibly furnished indoor scenes, we first extract layout information as prior, and then train a layout-constraint model. Afterwards, we construct an energy function which describes the reasonableness for indoor functional areas layout by combining the layout-constraint model with hard layout constraints. Finally, the energy function is optimized by stochastic optimization. Experimental results show that our method can generate reasonable furniture layout in complicated indoor space without specifying constraints for layout beforehand.