Annotate: AI-Accelerate Labeling and Dataset Management
Learn about cutting edge tools for preparing and training custom AI models like the one seen above.
For construction firms, estimating the cost of a project is a time-consuming and resource-intensive process. Before deciding to accept a project, the firm must review the design and count each individual component in order to approximate the total cost.
Traditionally, this process has been performed manually. Professionals spend hours reviewing each CAD drawing and tallying up the various building materials required for the project, such as electrical components, fire safety equipment, and other items.
After observing this challenge in their own design and cost estimation process, a team of developers at Systems Nakashima, a global manufacturer and technology provider, decided to create software that would use AI to automate the process of reviewing CAD drawings and estimating costs.
“Imagine you work at a construction firm that reviews 30 projects each month. Your team spends hours completing a cost estimate, and oftentimes you decline the project,” said Shingo Matsuura of Systems Nakashima. “We wanted to automate this process for other companies. Based on our previous experience, we knew Roboflow could help us accelerate development of the visual AI necessary for the project.”
Matsuura, along with his colleagues Satou and Mokkou, had experience developing software for industrial processes, but decided to use Roboflow to create and deploy the AI model that would power their cost estimation software. With Roboflow, they developed a custom vision model capable of analyzing CAD drawings and recognizing the technical symbols and markings of various components, such as electrical and fire safety equipment.
“When teaching the model to recognize the marks for electrical equipment, we annotated a dataset of over 11,000 images,” Matsuura said. “Training models to recognize these and all the other symbols in blueprints was only possible with Roboflow.”
By using the software from Systems Nakashima, construction firms and electrical contractors can scan CAD drawings and export a list of all the components in a matter of seconds, allowing them to drastically reduce the time required to calculate the estimated costs of a project.
Behind the scenes, Roboflow's cloud inference service has allowed System Nakashima to deploy and run their computer vision models without building and managing costly infrastructure. This enabled the team at Systems Nakashima to focus on developing other critical aspects of the software, like the user experience and integrating with other software commonly used in the construction industry.
“In Japan, labor shortages are becoming an increasingly serious issue, making AI-driven efficiency improvements, such as those enabled by Roboflow, more essential than ever,” said Matsuura. “At Systems Nakashima, we aim to contribute to the growth of our industrial sector by streamlining and automating tasks like these with AI.”
Systems Nakashima Co., Ltd. was established in 1985 after separating from Nakashima Propeller Co., Ltd., a leading manufacturer of marine propulsion units. Leveraging its expertise in computer technology and system development, the company initially focused on creating CAD systems, starting with the "ANDES" general-purpose CAD system for propeller design.
Over the years, Systems Nakashima expanded its business into multiple industries, including civil engineering, construction, electricity, and water supply. Among its key products, "Sui Shou" has become a market leader in CAD systems for water supply and drainage applications.
The company continues to innovate by integrating advanced technologies, developing specialized solutions, and forming strategic partnerships to provide comprehensive system integration and IT solutions across various sectors.
