The School of Architecture congratulates PhD in Computational Design (PhD-CD) student Jinmo Rhee for receiving the 2020 Young CAADRIA Award for his promising research exploring Artificial Intelligence and Machine Learning techniques in architecture.
Young CAADRIA Awards are funded by The Association for Computer-Aided Architectural Design Research in Asia (CAADRIA), and given by a committee consisting jointly of CAADRIA, the Paper Selection Committee, and the Conference Host. The winner is decided after the acceptance of papers. Awards are applied to conference registration fees. Due to COVID-19, this year’s 25th International Conference of the Association for Computer-Aided Architectural Design Research in Asia will be held as a virtual conference from August 5-8, 2020.
Learn more about Rhee’s awarded work below:
The Potential of Learning-Based Systems in Space Planning: A Sample Experiment of Building Footprint Prediction by Deep Learning
We present a method for generating building geometry based on urban context and explore its potential to support building massing. For contextual geometry, we opted to investigate the building footprint, a main interface between urban and architectural forms. For this purpose, we collected GIS data of building footprints and geometries of parcels from Pittsburgh and created a large dataset of Diagrammatic Image Dataset (DID). We employed a modified version of a VGG neural network to model the relationship between (x) a diagrammatic image of a building parcel and context without the footprint, and (y) a quadrilateral representing the original footprint. The option for simple geometrical output enables direct integration with custom design workflows because it obviates image processing and increases training speed. Lastly, we explore a generative workflow for building massing that integrates contextual and programmatic data. After training the neural network with a curated dataset, it can suggest a contextual boundary for a new site. Then, using (the name of previously developed program omitted for blind review) to recommend massing alternatives based on the subtraction of voids inside the contextual boundary that satisfy design constraints and programmatic requirements.
Keywords: Deep Learning, Prediction, Building Footprint, Massing
The paper was co-authored by CMU School of Architecture PhD in Computational Design candidate Pedro Veloso and Professor Ramesh Krishnamurti.