Dec. 6, 2021

Computer vision research's bottleneck can often be the availability and annotation of image data. We present a new tool for using deep learning models to automatically annotate images, and allowing specialized labeling tools to be integrated as plugins for niche scientific research aspects.

We demonstrate how a model zoo can be used to quickly annotate datasets in a scientific domain such as microscopy or laboratory automation, and then demonstrate how to write and integrate a new tool for an uncommon annotation type required within the research domain.

We will proceed to train a small model using the automatically generated annotations and present the results.

Graphotate's launch is December 3, 2018 and has been almost a year in the making. This will be the first public demo of the tool.