An Improved CGAN-Based pix۲pix Approach for Building Facade Prediction (An inovative technology)

Nowadays, the application of artificial intelligence (AI) has increased in many branches of engineering sciences. Use of AI in fields of engineering increases speed and quality of executive operations. One of the fields that used AI for more performance's exploitation is architectural engineering. Using AI into Building facades can change the efficiency and implementation of existing activities in this field. On the other hand, by using AI in building facades design, architects and engineers can implement the design process more simply, faster and more accurately. Also, with the use of AI as a required tool, the functional quality of construction will be improved in terms of criteria such as cost, versatility, and structural quality. In this research, the goal is to predict the actual view from an existing basic map. In order to produce such a view using deep learning. In the algorithm, used a conditional generative adversarial network (C-GAN) with the ability image to image translate. In the proposed method a Unet architecture was used in the generator part and ca deep convolution network used in the discriminator part. The results of the implementation showed that the application of the proposed algorithm has the ability to display the general outline of the initial map with very good accuracy.

Keywords:

Architectural Design , Building Design , Conditional Generative Adversarial Network , Image Translation