Practicing AI utilization with NVIDIA GPU as a platform

・Development of image analysis using cameras in two months from consultation to installation

・Realization of visualization of the number of users and congestion status

・ Macnica 's support will also help apply AI to the entire real estate business

Mitsui Fudosan Co., Ltd.

The group develops from design to operation management of various facilities and buildings such as office buildings, commercial facilities, hotels and resorts, condominiums, and custom-built houses, and boasts one of the highest sales in the real estate industry. With the Group's long-term management policy "VISION 2025" in mind, we have positioned "utilizing technology to innovate the real estate industry itself" as an important company-wide policy, and are working to innovate the real estate industry itself and work on further globalization. promoting.

 

Company HP: https://www.mitsuifudosan.co.jp/

Background: It was essential for new challenges that contribute to aging

Mr. Akira Sasaki, manager of the Kansai Branch Business Department 2 Business Group, which manages the Nakanoshima Mitsui Building, started the project because he wanted to create a building that "gets better with age," increasing its value over time. It was the trigger.

“It has been 17 years since the construction of the Nakanoshima Mitsui Building was completed, and we have tried to bolster it several times during that time. (Mr. Sasaki)

One of the ideas that came up was the use of cutting-edge technologies such as AI, but Mr. Ryo Yoshioka of the same group says that the use of AI was not a premise from the beginning.

“We did two renovations in a short period of time, but we weren’t able to learn enough, so we needed some sort of environment that would allow us to visualize the current situation. Of course, we also wanted to establish a process that would allow us to flexibly modify the hardware according to the environment." (Mr. Yoshioka)

When the company sought advice from the Venture Co-Creation Division, which promotes the use of technology within the company, they received advice on new technologies such as image analysis and AI, and as part of that advice they were introduced to NVIDIA, which has powerful GPUs suitable for building a foundation for IVA (Intelligent Video Analytics). NVIDIA then introduced the company to Macnica.

While consulting with Macnica, they introduced a solution that combined network cameras and real-time image analysis using AI as a system that could be used to set up an environment in a short period of time and also analyze spatial information such as facility usage status. .

“Actually, many of our business partners have proposed various solutions, such as systems that use sensors, but what we wanted to do the most was to improve the space itself. What we thought would be useful for our policies was a solution using a camera, and we thought that it could be used for future predictions, including analysis by AI." (Mr. Yoshioka)

Installation: 2 months from consultation to installation, preparation of environment for visualization and image analysis using cameras

Yoshioka recalls that Macnica 's proposal was shocking:

“I know many partners who can help me realize what I want to do, and I was able to expand the world from my narrow perspective.I was shocked at what a technology trading company was like.”

As a result, a solution proposed by Macnica using the NVIDIA GPU deep learning platform and Optim, which has a rich track record in image analysis algorithms, was adopted, but one of the major points was that it could be implemented in a short period of time. Ta.

“Originally, we had decided to renovate the cafeteria at the Nakanoshima Mitsui Building, but we started looking into a camera-based solution about three months before it opened. (Sasaki)

In fact, it took only about two months to set up the camera, prepare the infrastructure for image analysis, make adjustments, and test operation. Mr. Sasaki highly appreciated the fact that the demonstration experiment could be started in a short period of time.

42 cameras are installed across the three floors, and OPTiM AI Camera for Office Building provided by OPTiM on the NVIDIA GPU deep learning platform is used to visualize the number of users and congestion status. there is

効果：具体的な成果につなげる活動と不動産全体へのAI活用に期待

Although it is currently in the PoC stage (*6 months from October 2019 to the end of March 2020) toward full-scale introduction, it is possible to count people with high accuracy, so it is possible to analyze measures to promote usage. An effective environment has been created. In addition, when we analyzed the usage of the seats that were originally set up for one person, we found that there were many cases where two people were using the seats. . Furthermore, with regard to visualization of congestion status, it has become possible to grasp the congestion status by presenting it in real time on the website for tenants.

We were amateurs in everything but building management, and Macnica was the one who connected us with the various technologies and languages in an easy-to-understand way. They understood what we wanted and put it into a system that allowed us to realize the project in a short period of time and take the first step toward achieving superiority over time," said Mr. Yoshioka.

Mr. Daichi Kamo, a member of the same group who has been participating in the project since 2019, added, ``As Yoshioka progressed the project, I often saw him communicating closely with Macnica personnel.''

In the future, it will be necessary to analyze the information obtained from the camera and connect it to concrete results.

“From the standpoint of promoting the business, it is necessary to draw a picture of the future while showing concrete results. Full-fledged analysis starts now.” (Mr. Sasaki)

He added, ``Currently, we are only deploying this on some floors of the Nakanoshima Mitsui Building, but by spreading this initiative within the company and leveraging our knowledge of commercial relocations that attract unspecified large numbers of people, we will be able to expand this initiative within 10 years. "We want to make image analysis using cameras a common practice in the future. To that end, we would like to utilize Macnica 's wide range of knowledge and technology to improve the convenience and comfort of facilities," said Mr. Yoshioka. talk.

Furthermore, the company is looking forward to Macnica 's proposals as a technology trading company beyond office buildings, such as systems that contribute to creating attractive experiences that motivate visitors to visit commercial facilities.

Mitsui Fudosan will continue to take on the challenge of applying new technologies to its business, such as utilizing AI technology based on NVIDIA GPUs, and will continue to work with Macnica to bring about change in the real estate industry using new technologies.

・Company names and product names mentioned are trademarks or registered trademarks of their respective companies.

・Trademarks (TM, ®) are not necessarily added to system names and product names on this page.

・The information on this page is current at the time of the interview. The contents of this page are subject to change without notice.

Click here for other NVIDIA product introduction examples