Layout views

There are a few different ways you can view your project’s items through the Explore tab: Grid View, Instance View, List View, and Zoom View.

From each view type, there are a number of bulk actions you can take. The availability of these actions depend on the view and project type. Learn more.

Grid View

This view method is available for Image and Text projects. It allows you to see the items in your project as thumbnails. By hovering over them, you can approve or disapprove instances with ease. You can make use of various filter tools, and categorize your items into Subsets.

Instance View

This view is unique to Image projects. Here, you can see thumbnails of all of your instances across every item in your project. By hovering over them, you can approve or disapprove instances with ease. You can make use of various filter tools and bulk actions on instances.

List View

This layout is unique to Video and Multimodal projects. You can see all of your items displayed as a list, make use of various filter tools, and categorize your items into Subsets. For Multimodal projects, you can also customize your list to view specific item parameters based on the selected criteria. Learn more.

Zoom View

When viewing projects in the Explore tab, you can click on any item to look at it in Zoom View. It is a simple window that gives you a quick and easy way to examine annotations thoroughly, approve or disapprove items, or leave comments on them to provide feedback. You can categorize items into Subsets.

Through this view, Project Admins can assign or change the item’s assigned contributors. For Multimodal projects only, you can add categories to items, and move them to different folders.

If you still want to examine the item further, you can click Open in editor so that you can see the item in full view. You can also access an item in the editor by Control / Command + clicking it from the Explore tab.

📘
This feature is currently only available for Image and Multimodal projects.

🚧
For Multimodal projects, you can only add one comment to each component.

Filter tools

From the top panel, there are a few ways you can filter and manage how you view your dataset.

Filter by folder - You can select a project folder from the dropdown list to view items from that folder specifically. By selecting a folder, you can also move or copy items from that folder to another in bulk.
Filter by subset - You can select a subset from the dropdown list to view all items within that subset. By selecting a subset, you can also move or copy items from that subset to another in bulk.
Natural language search - From the top panel, you can make use of the Natural Language Search feature to search for specific items that match the given key words.
Queries - You can type JSON Queries into the query bar in the top panel to curate specific results within your project.
Instance View - For instance view, there is a list filters that you can use to manage the instances that you see.

Explore settings

In Explore, there are various settings you can configure to view your dataset with more specific criteria.

Columns

This section can be found in the left panel of Grid View. It allows you to choose how many columns you want to display your items in to best accommodate you. You can choose between 1 - 5 columns.

Display

This section can be found in the left panel of Grid View and Instance View. It allows you to fine-tune your filtering and curation in a multitude of ways. Some of the filters listed below may only be available for specific projects:

Opacity lets you change the opacity of instances on each item that you can see in the Explore tab.
Item Information has three options you can enable:
- Information - This'll enable tooltip icons on the thumbnail that give you the item’s name, QA and Annotator, item status and approval status.
- Classes - This'll show you the classes assigned to instances.
- Attributes - This'll show you the attributes assigned to instances.
- Point Labels - This’ll show you the point labels assigned to instances.

Filter

This tab can be found in the left panel of the Instance View. Here, you'll be able to select what you want your data to be filtered by, and these filters are categorized as follows:

Class Name allows you to filter the visible instances based on specific class names.
Approval status lets you see which instances are approved, disapproved or have no approval status at all.
Instance type lets you filter by annotation tools.
Group ID lets you filter by instance group IDs.
Creation type lets you see the instances that were created manually or by pre-annotation.
Created At allows you to see which instances were created between a given range of time.
Updated At allows you to see which instances were updated between a given range of time.

Resync

Whenever an item is being worked on in the editor, that data is automatically updated in Explore. Resynchronization, or Resync, is a way for you to give your dataset a manual update.

Sometimes, there might be a synchronization issue between the changes made within editors, and the visible dataset. A simple refresh of the page could help, but there are a few cases in which you’ll have to resync your dataset:

For all project types - When new features are released, you can resync your dataset to apply any necessary changes for those new features to be implemented and usable. You’ll be notified if a resync is required in this case.
For all project types, except Multimodal - If you’ve deleted a class through Project Setup or with SDK, and that class has been used to classify instances in a project, those instances will become unclassified. Resyncing your dataset will make sure these instances appear correctly in Explore.
Multimodal projects - If you’ve deleted a category through the project’s settings, and that category has been used to categorize items in a project, those items will become uncategorized. Resyncing your dataset will make sure these items appear correctly in Explore.
For Image and Text projects - When generating embeddings for items for features such as Find Similar or Natural Language Search, if the item batch number is too large, a few items might get skipped over. Resyncing helps to go over your dataset in this case and find any items that were missed during embedding generation so they can also be used with these features.