Automation in Data Management and Data Labeling

The Data Exchange Podcast: Hyun Kim on building automation and DataOps tools to help companies unlock computer vision data.

SubscribeApple • Android • Spotify • Stitcher • Google • RSS.

This week’s guest1 is Hyun Kim, co-founder and CEO of Superb AI, a startup building tools to help companies manage data across the entire machine learning application lifecycle. This includes tools to label, store, and monitor data assets that power all computer vision applications. We also discussed emerging trends in machine learning and AI including synthetic data, reinforcement learning, and self-supervised learning.

Superb AI is assembling a suite of tools for automating many aspects of DataOps for computer vision, from labeling, data management, auditing and QA,  monitoring, and governance.

    ❛ “Auto labeling” is a name for our automated data labeling feature. It basically gives users a pre-trained model that they can use to label their data. In addition to the “auto label” model, we provide a custom model label. Let’s say a user has a very niche data set, maybe from a specific manufacturing pipeline. And this user wants to have our model be fine tuned on their data. So what such users will do is they’ll manually label maybe a few – maybe 100 – images, and use these labeled images to finetune our model using just a few clicks. They hit a button and wait around 30 minutes to an hour, and our platform will produce a model that’s fine tuned to the client’s data. And from that point on, they can use the new updated model to label the next batch of their data set.

Download a complete transcript of this episode by filling out the form below:

Related content and resources:

[1] This post and episode are part of a collaboration between Gradient Flow and Superb AI. See our statement of editorial independence.

[Image from Peggy und Marco Lachmann-Anke on Pixabay.]