September 2021 update

September 2021 Update

September 21, 2021
A series of cards from V7 Darwin

Train a Classification Model

Use your tagged data to easily train a classification model in a few clicks. Add your model to a workflow, and automatically tag thousands of files in minutes.

Create a Workflow with Consensus Stages (Beta)

Reach out to us to test Consensus Stages in beta and have multiple annotators label the same files in parallel.

Add Attributes to Tags

Add an additional layer of data to your tags - all tag classes now support the addition of Attributes.

SSO Support

We now support SAML Single Sign On for teams using Azure as part of our Enterprise plan.

Easily Cancel or Complete Tablet Annotations

A new Cancel/Complete pill makes completing polygons or polylines easier than ever when annotating on a tablet or mobile device. 


🛠 Removed the option to turn 2FA on during signup, reducing the chance of users getting locked out right away.
🛠 Fixed an issue that caused some classes to be removed from their datasets. 
🛠 Fixed the counts on the Neural Networks page which indicate which items should be trained on. 
🛠 Fixed a pagination issue when displaying classes that can be trained on in the Neural Networks page.
🛠 Fixed another pagination issue that prevented easy scrolling through team members in the Settings menu.
🛠 Fixed an issue with upload signing when importing files from external S3 buckets.
🛠 Fixed an issue with item counts not taking certain filters into consideration in the Data page of a dataset.
🛠 Fixed an issue that prevented an item's class from being changed when it contained a subannotation like an Instance ID or Attribute.
🛠 Fixed an issue that prevented Undo and Redo when using Auto Annotate.
🛠 Fixed an issue that allowed changing an item's class to change its class type.
🛠 Fixed an issue to correct the direction in which attributes interpolate when annotating a video file.
🛠Pushed a new version of our SDK that fixes concurrency issues that would crash imports with many files.
🛠 Fixed an issue that prevented model training from taking place succesfully when any file in the training dataset contained a space in its name.

Performance Improvements

🚀 Loading and counting items is now way faster in datasets that contain a large number of videos.