Japanese real estate technology company GA technologies (main office: Minato-ku, Tokyo, securities code: 3491) is pleased to announce the start of a joint research project in the ConTech domain. This research will be conducted in collaboration with the Furukawa Lab (*1), a research group led by Associate Professor Yasutaka Furukawa, a leading figure in the field of image analysis and 3D modeling in the real estate and architectural domains. Professor Furukawa is associated with the GrUVi Lab and the VML Lab at Simon Fraser University (Canada). The starting point for this project is to improve precision for a deep learning (AI) floorplan scanning system, BLUEPRINT by RENOSY, which was released as a beta version in July 2019.
– Joint research with Furukawa Lab, a leading research group in image analysis and 3D modeling in the real estate industry
– The first step of the research will be to improve precision for an AI floorplan scanning system, BLUEPRINT by RENOSY
– Evaluating potential use in 3D modeling and Building Information Modeling (BIM), and the provision of building data to the real estate market
– Overview of research:
BLUEPRINT by RENOSY is a semi-automation tool that generates CAD files from floor plans for condominium renovations. Image recognition technology to detect walls and windows as well as an optimization process to ensure consistency in the position of each element is vital for this process. By applying the Furukawa Lab’s advanced knowledge of image recognition and optimization, GA
technologies aims to increase the precision of each process and develop a tool that minimizes human input.
Boasting a market scale of 50 trillion yen (*2), the Japanese construction industry is second only to the automotive industry. At the same time, the construction industry lags in the adoption of technology promoting workflow optimization, resulting in over 300 more hours of overtime than the average across all industries. (*3)
GA technologies developed BLUEPRINT by RENOSY to respond to the urgent need for the adoption of technology in the construction industry. To further improve the precision of BLUEPRINT by RENOSY and expand its adoption, the company will be launching joint research projects in the ConTech domain. The first of these projects is the collaboration with the Furukawa Lab.
The first aim for this research project will be to reduce the time required for BLUEPRINT by RENOSY to generate the CAD diagram by 90%. GA technologies then aims to apply its research to 3D modeling using scanned diagrams and eventually work towards its use in BIM.
Comments on research motivation and goals:
– Comments from Professor Yasutaka Furukawa of the GrUVi Lab and the VML Lab at Simon Fraser University
“With the remarkable advancement of deep learning and other forms of machine learning technology, the automation of image analysis is becoming the norm in the real estate and construction industries. My lab has contributed to the commercialization of this technology in the real estate industry by being at the forefront of research in areas such as the automated conversion of floorplan images to CAD files and the automated generation of floorplans using depth images. Through joint research with GA technologies, I want to take the first step towards expanding the scope of application for this image analysis and restoration technology from a focus on floorplans, which are mainly used as visual references for customers, to use in Building Information Modeling such as building schematics that can be used in actual construction and renovation work.”
– Comments from GA technologies President and CEO Ryo Higuchi “While Japan continues to promote workstyle reform (*4), the construction and real estate industries still face numerous issues related to the reliance on labor-intensive processes. Thus far, we have provided various products aimed at improving the efficiency in processes in the real estate and construction industries. We are incredibly honored to be given the opportunity to work with the members of the Furukawa Lab, which is internationally recognized for its research in the fields of image analysis and artificial intelligence in the real estate and construction industries. We hope that by applying the results of this joint research, we will be able to further enhance our services, and contribute to the promotion of workstyle reform for the entire industry.”
(*1) The Furukawa Lab is internationally recognized for its numerous achievements in research related to image analysis and 3D modeling in the real estate and construction industries. Many of its research results have been recognized at world-leading conferences on image analysis and artificial
intelligence, including ICCV and CVPR.
(*2) Source: MLIT Estimate of Construction Investments FY2018
(*3) Source: MLIT Construction Industry Workstyle Reform Acceleration Program (FY2018)
(*4) Workstyle reform: The “Work Style Reform Law” was passed on June 29, 2018, by the Abe administration to improve the Japanese working environment.
– About BLUEPRINT by RENOSY
A system that uses deep learning (AI) to read floorplans and automatically generate digital data (CAD data) for existing buildings requiring renovation design.
– Overview of GA technologies
Company name: GA technologies Co., Ltd.
Representative: Ryo Higuchi, President and CEO
Main office: Sumitomo Fudosan Roppongi Grand Tower 40F 3-2-1 Roppongi,
Established: March 2013
Capital: 1,168,698,450 Japanese yen (As of May 31, 2020)
– Operation of “RENOSY”, a comprehensive PropTech brand (Real estate information media, real estate brokerage, real estate sales, design and construction, real estate management)
– Developing BtoB PropTech SaaS Products.
– Research on real estate big data using AI
– Operation of “Shenjumiaosuan”, a Japanese real estate platform for investors in Greater China and other international PropTech business.
Group Companies: ITANDI Inc., Modern Standard Co., Ltd. and other 3 companies.
– Overview of Furukawa Lab at Simon Fraser University
Lab website: https://www.cs.sfu.ca/~furukawa/
Simon Fraser University website: https://www.sfu.ca/
Director: Associate Professor Yasutaka Furukawa
Main research themes: Research in reconstructing 3D models of historical buildings using large volumes of photographs taken by tourists, research in establishing 3D models by analyzing images of room floorplans, etc. Professor Furukawa has a remarkable record of success in research related to image processing in the real estate and construction domains.