Automatic Indoor Space Layout Design Based on Deep Reinforcement Learning

With the growth of homeownership, room design also affects the demand for interior decorators and those difficult to find interior designers because people always see new designs and challenges keep changing the trends. Due to that reason, this research proposes a spatial area layout algorithm based on Deep Reinforcement Learning (DRL) of deep learning. It offers a two phase approach that describes room initially and wall, as well as additionally offers a big-scale dataset referred to as RPLAN, which incorporates further than 80,000 indoors format plans from real residential buildings. The proposed approach use both supervised and unsupervised Graph Neural Networks (GNNs) to construct plans, deploying such approaches. An amount of verifiable evidence has been presented to support the fact that this method works in arrangement of interiors. The prediction starts from the living room and is completed by automatic layout iteration of the interior areas. The similitude when evaluated by skilled interior designers are noticeable when their floor plans are compared.