At SuperAnnotate, we design new features, fine-tune the existing tools, and integrate new functionalities regularly to provide the best possible experience for you. In this article, we’ll list important updates on the current cycle.
Release date: March 20, 2022
In this cycle, we intend to ensure smoother navigation across the platform by presenting a new projects page with additional project management actions. We’re also introducing a pixel-editor-specific approach in the Vector Editor, a GCP integration, SDK functions and much more. To meet our objectives, we are releasing the following updates:
- Project list redesign
- GCP integration
- Query Search for Video Projects
- Subtract Polygon: Improved vector experience
- Vector Editor template with workflow connections
- Nested classification: Tagging with attributes
- Integrations in SDK
Project list updates
Project view improvement: The current release provides an intuitive overview of all projects on one page instead of having them displayed on the secondary navigation bar.
Project creation flow improvement: We also streamlined the project creation process by eliminating pop-ups. You can now create new projects by selecting the type of the project from the drop-down menu.
Project actions redesign: We added two action columns:
- Track your project progress more effectively by updating its status. Select from the following status update options:
- Not started
- In progress
- On hold
2. Now, you have quicker access to project actions, such as Rename, Pin/Unpin, Delete. How is this helpful?
- You can edit the project name with fewer steps without entering the Project Settings.
- Pinning or unpinning projects from the list will allow you to filter and access high-priority projects faster and with ease.
Breadcrumb inside the project: Navigate through projects and folders through a breadcrumb; You can switch views or change your location/path on the platform with a few clicks.
In our current release, we introduce a Google Cloud Platform (GCP) integration: You can add items (images, videos, documents) from your GCP storage to SuperAnnotate, providing us with read-only access to your data. That means your items are securely stored in your private storage and are only displayed in SuperAnnotate.
To enable GCP integration, go to the integrations page, create a service account, and grant corresponding permissions. Then, upload the service account’s key, and validate integration.
When uploading items with GCP, the bucket’s folder structure and its data will be replicated in SuperAnnotate’s two-level folder structure, eliminating the manual one-by-one data transfer.
Query Search for Video Projects
By now, you should be familiar with our Query Search for image projects: We’re launching one for Video Projects as well. SuperAnnotate’s query layout inside the Video Projects will simplify managing, filtering, and finding videos based on specific criteria. You can either type queries in the bar in a SuperAnnotate Query Language (SAQuL) format or use autosuggestions.
Subtract Polygon: Improved vector experience
Moving on, SuperAnnotate developed a solution to help you annotate objects with difficult structures. Our new Subtract Polygon feature offers a pixel-editor-specific approach to annotation. Now you can remove part of a Polygon by drawing a child polygon over the area you want to extract from the existing (parent) polygon.
Vector Editor template with workflow connections
With our new Vector Editor template with workflow connections, you can perform skeleton and other template-related tasks much faster. The feature also allows you to:
- Annotate the occluded points of the template
- Assign a class/attribute to each point
- Delete a point from the template
Use the Point tool to set up connections between the Workflow steps and accelerate your project speed. The steps that are connected to each other will have their corresponding keypoints connected and will belong to the same group.
Nested classification: Tagging with attributes
Remember nested tags in our Vector Editor? We’re introducing them in the Document Editor as well. Now you’ll notice two types of tags: Tags (that are considered instances) and Free Text Tags. The setup in the Class tab is the same as in Vector Projects.
Integrations in SDK
Finally, with this release, we’re adding get_integrations() and attach_items_from_integrated_storage() functions that allow getting information about existing integrations and attaching items automatically from the integrated storage, thus, fully automating the data import.
You can also import active learning priority scores to the SDK through a simple script of Python commands. Check out more details about updates in our Python SDK changelog.
If you have feedback or suggestions to help us further improve the platform, let us know.