Scale and nuTonomy open-source massive AI dataset for self-driving cars

Scale Labs Inc. and Aptiv PLC’s nuTonomy group, two influential players in the autonomous vehicle ecosystem, today open-sourced a massive research dataset designed to aid self-driving car initiatives.

Autonomous vehicles rely on artificial intelligence models to make navigation decisions. Those AI models, in turn, must be trained with large amounts of sample information to achieve the necessary accuracy, which is where the new dataset comes into the picture.

Scale and nuTonomy are touting the library as the largest and most detailed among the self-driving datasets on the market. According to the companies, this includes DeepDrive, a dataset developed by researchers from the University of California at Berkeley that’s considered the go-to choice for autonomous driving projects.  

DeepDrive is comprised of 100,000 video sequences and 100,000 images taken during drives on public roads. NuScenes, as Scale and nuTonomy are calling their dataset, contains only 1,000 20-second clips but nearly 1.4 million photos along with 400,000 three-dimensional scans taken using LIDAR sensors. Topping it all off are 400,000 bounding boxes extracted from the images that highlight objects of interest.

The data collection effort was led by nuTonomy. Originally founded in 2013 as a spinoff from the Massachusetts Institute of Technology, the group has developed a “full-stack” software platform for autonomous driving. It was acquired by Aptiv last October, back when the auto parts giant was still known as Delphi, in a deal worth over $400 million.

Scale, in turn, has raised $22.6 million from investors including Accel and provides a data annotation service specifically geared towards AI training data. The platform was used to label the photos and videos that make up NuScenes. Tags explaining the meaning of records help machine learning models learn more efficiently during the training phase of development.

Many of the companies in the self-driving car ecosystem that could potentially use nuScenes already work with either nuTonomy or Scale. The latter startup alone counts more than a dozen key players as customers, including Lyft Inc., Drive.ai Inc. and General Motors Co.’s multibillion-dollar Cruise Automation unit.

Photo: NuTonomy

A message from John Furrier, co-founder of SiliconANGLE:

Show your support for our mission by joining our Cube Club and Cube Event Community of experts. Join the community that includes Amazon Web Services and Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger and many more luminaries and experts.

Join Our Community 

Click here to join the free and open Startup Showcase event.

“TheCUBE is part of re:Invent, you know, you guys really are a part of the event and we really appreciate your coming here and I know people appreciate the content you create as well” – Andy Jassy

We really want to hear from you, and we’re looking forward to seeing you at the event and in theCUBE Club.

Click here to join the free and open Startup Showcase event.