+82 10 9187 2216

Architectural Diagrammatic Image Dataset for Learning

  • machine learning, AI, space syntax
  • Designed by Jinmo Rhee
  • Collaboration with Pedro Veloso

The project is the first step in bringing AI technology to architecture, an experiment on how machines can understand the architectural context. Previously, in other areas, satellite images and pictures of real objects were studied by Convolution Neural Network. At this time, the unwanted information was regarded as noise. We apply this to make a diagram showing the relation and experiment on whether artificial intelligence can understand it. This study is a research that shows the possibility of constructing a system that can establish the relationship that the architect wants to analyze beyond a simple relationship, and to deduce the relationship that human beings can not recognize or express in a simple function. The actual "artificial intelligence can recognize this diagram" proved the possibility of understanding the architectural context. This is the first step in designing artificial intelligence.

Process of Architectural Diagrammtic Image Dataset Generation

related projects

next work


School with enviromental techinics. In the images were intended to show the pleasant spaces. The image shows cafeteria space of the school.

another work


Housing Project. The town has some outer spaces with a hierarchy. Inside looks of hhe city, which is already existed, is ...