Generate Large, Synthetic, Annotated, & Photorealistic Datasets

Our goal is to make it quick and easy to generate a lot of training data for computer vision object recognition / semantic segmentation / instance segmentation.

Please read our beta announcement blog post.

We'd also love to help on your project if we can - email me at matt@greppy.co or reach out to me on LinkedIn to connect.

 

Quick Overview

In a nutshell, you (1) create a type of thing "Thing" you want to recognize, like a bottle or coffee machine. (2) Then, you upload 3D/CAD models of different variations of that thing you'll come across.

(3) you attach "Materials" to each 3D/CAD model you upload, and (4) specify its parts that you want to recognize as well.

Finally (5), select the "Camera" you want to generate the dataset for, and (6) get your "Dataset" rendered for you!