What do we do?

Image by Lucas George Wendt

Data Generation.

Data Labeling.

Augmentation.

Synthetic Data is Data 2.0

 

If you are training computer vision models to solve complex problems, you need synthetic data. Synthetic data is the future of all machine learning training data.

  • More Data:​ ​

We use CGI and 3D rendering engines and leverage the power of cloud computing and ML to generate unlimited data sets.​

  • Better Data:

Automatic, pixel perfect data labeling. ​We can target specific, hard-to-get edge cases. Our data is self-improving. Auto-ML for data!

  • Built-In Privacy:​

U​nbiased, diverse data free of any privacy issues.  ​Our data can equally represent all ethnicities, races, and genders.

 

Synthetic data vs rudimentary labels.PNG

What We Offer

Zumo Labs​ offers both custom synthetic data solutions and open source developer tools for synthetic data generation. 

Curious, want to play around with our tools? Check out ZPY.

 

Want us to take from here? Reach out to us for custom synthetic data solutions. How does it work?

  • We consider your use case and the ML models that you are using, training, and/or improving.

 

  • We create a simulation or scene.

  • We generate data (images, animations, videos, you name it) for you. We have full state knowledge of the scene, meaning we provide:

  • Full key-point skeleton

  • Normals

  • Depth

  • Stereo

  • Class and instance segmentation

  • Multiple camera angles

  • Instance tracking​

  • Occluded key-points

  • 6D pose

  • And much more