Rendered.ai makes AI training easier for developers, data scientists, and others with a platform-as-a-service for synthetic data generation (SDG).
Training computer vision AI models requires large, high-quality, diverse, and unbiased datasets. These can be hard to come by and costly, especially as the demand for both AI and AI increases.
The Rendered.ai Platform as a Service helps solve this problem by generating physically accurate synthetic data (created from 3D simulations) to train computer vision models.
Rendered.ai Founder and CEO Nathan Kundtz said, “Real-world data often cannot capture all the possible scenarios and edge cases needed to generalize an AI model. The SDGs are key for AI and machine learning engineers.” , based in Bellevue, a suburb of Seattle, Washington.
Rendered.ai, a member of the NVIDIA Inception program for cutting-edge startups, has been integrated into the NVIDIA Omniverse Replicator platform, a core extension of the Omniverse platform for developing and deploying industrial metaverse applications.
Omniverse Replicator enables developers to generate labeled synthetic data for many applications such as visual inspection, robotics, and autonomous driving. It is built on open standards for 3D workflows such as Universal Scene Description (“OpenUSD”), Material Definition Language (MDL), and PhysX.
Composite images generated by Rendered.ai are used to model landscapes and vegetation in virtual worlds, detect objects in satellite imagery, and even test the viability of human oocytes and egg cells.
Rendered.ai leverages Omniverse Replicator’s RTX acceleration capabilities (ray tracing, domain randomization, multi-sensor simulation, etc.) to enable computer vision engineers, data scientists, and others to perform compositing through a simple web interface. Data can be generated quickly and easily. Cloud.
“The data needed to train an AI is really a key driver of AI performance,” says Kundtz. “The integration of Omniverse Replicator into his Rendered.ai brings a new level of ease and efficiency for users to harness synthetic data to train bigger and better AI models for applications across industries. will be possible.”
Rendered.ai will be demonstrating platform integration with Omniverse Replicator at the Computer Vision and Pattern Recognition (CVPR) conference June 18-22 in Vancouver, Canada.
Generate synthetic data in the cloud
Rendered.ai, now available through the AWS Marketplace, brings a collaborative web interface to the cloud for developers and teams to design SDG applications, making it easy for computer vision engineers and data scientists to configure.
It’s a one-stop shop where people can share workspaces containing SDG datasets, tasks, graphs, etc. all through a web browser.
Omniverse Replicator tutorials, samples, and other 3D assets are now readily available on the Rendered.ai channel running on AWS infrastructure. This gives developers a starting point to build their own SDG functionality on the cloud. It also makes the process especially seamless for Omniverse Replicator users who are already familiar with these tools.
According to Kundtz, Rendered.ai is focused on providing physically accurate synthetic data through its platform, which allows AI systems to build on how real-world physics processes work. new information can be introduced into
“In the future, every company using AI will have a synthetic data engineering team for physics-based modeling and domain-specific AI model building,” he added. “Since complex AI tasks require training on large amounts of data covering different scenarios, simulating something once is often not enough. Synthetic data is key here. ”
Hear more from Kundtz on the NVIDIA AI podcast.
Click here for details Omniverse Replicatorvisit Rendered.ai at CVPR Booth 1125 and Demo of this platform integration.
Explore new cloud capabilities. Rendered.ai site Sign up for a 30-day trial of a Rendered.ai developer account using the content code “OVDEMO”. This will give you access to his new Rendered.ai channel in Omniverse.
Featured image courtesy of Rendered.ai.
