Matterport’s 3-D media platform lets anyone turn physical spaces into immersive digital twins. To optimize machine learning algorithms and the value of the company’s spatial data library, CloudFactory provided quality-assured image annotation and classification.
Matterport’s 3-D media platform makes it easy for anyone—from individuals to Fortune 500 companies—to turn a physical space into an immersive digital twin. This innovation is made possible through Cortex, Matterport’s powerful AI technology.
As a deep learning neural network, it automatically stitches images together from a wide variety of capture devices—including LiDAR and 360-degree cameras, drones, and smartphones—to construct dimensionally accurate 3-D models that can be customized and shared. Billions of square feet have already been captured using Matterport.
From real estate photography and construction documentation to insurance adjusting and rental management and more, Matterport helps reduce the expenses, time commitments, and inefficiencies associated with managing the lifecycle of a building. The spatial data company’s goal is to digitize and index the built world by capturing interactive 3-D models that can be used to design, build, operate, promote, and understand any space.
While Matterport focuses on delivering the highest value possible to its customers through the creation of cutting-edge technological solutions, it also needs to maximize the value of its extensive database and huge library of images. The company owns the largest spatial data library in the world (and it gets bigger every day).
“Machine learning lies at the heart of our efforts to generate core analytics and insights for our customers,” explains Matterport Senior Deep Learning Engineer Sachal Dhillon. “This includes everything from constructing digital twins to identifying notable points of interest within them and much more.”
Its vast spatial data library enables the development of new neural network architectures and the ability to evaluate and optimize them against user behavior and real-world data in millions of situations. To create robust, highly accurate 3-D digital twins, it’s vital for Matterport to continuously optimize these neural networks and machine learning algorithms so that essential knowledge can be extracted from the growing volumes of data.
Due to the sheer number of images captured, the spatial data company needed a partner that could provide image annotation and classification services for its database of 3-D panoramic scans. Instead of spending time on data labeling, the company wanted its in-house resources to focus on development.
As Matterport continues on its trajectory of rapid and exciting new growth, partners like CloudFactory give us the ability to move quickly and unlock the full potential of our digital solutions through the delivery of high-quality data annotation at scale.
To help Matterport optimize machine learning algorithms, CloudFactory performed image annotation and classification across Matterport’s database of 3-D panoramic scans, including managed image segmentation services for a large breadth of ontologies and object types. Scanned spaces range from apartment buildings and schools to historic sites and luxury yachts.
Matterport supplies sets of images to CloudFactory, along with specific processing instructions and details. From there, completed and quality-assured annotations are returned. This information helps the company build tools for its customers to unlock valuable spatial data insights and analytics.
“The managed quality assurance offered as part of CloudFactory’s services has meant that we could rapidly iterate on and explore new data annotation projects without needing to build out new workflows every time,” says Dhillon. “This has been critical in allowing us to get new products and technologies out to market as fast as possible.”
By augmenting machine learning capabilities and removing the burden of image annotation from Matterport’s workforce, CloudFactory helps the spatial data company remain focused on engineering and developing cutting-edge spatial data technology solutions to take buildings online.
“The CloudFactory team has been a pleasure to work with,” says Dhillon. “We especially appreciate the proactive approach it has had in ensuring the delivery of high-quality results. Whenever an opportunity has arisen to further optimize the data annotation process or hammer down any points of ambiguity, we have been able to rely on CloudFactory’s experienced team of annotators.”
We have 10+ years of experience helping our clients focus on what matters most. See what we can do to help your business.
Geckomatics turned to CloudFactory to help scale their cost-effective solution to more customers around the world.
Geospatial mapping company turned to CloudFactory to take over its data labeling function and scale operations.