In the right-hand side of the Data page of any dataset, you'll see a breakdown of all images or videos by their state. Click the three dots at the bottom of the list to reveal some of the less common states, and select any item to filter for the images or videos in that stage of your workflow.
Each of the workflow stages you've added under the Settings page of your dataset will appear above the filters list. Click on any of the icons to quickly pull up the images or videos in one or more workflow stages.
Filter by annotator to pull up any of the images or videos that have been worked on by one or more members of your team. This will reveal all of the items that are either assigned to the annotator, or have been worked on by them in previous stages.
Filters carry over between the full view of your dataset in the Data page, and the work view (the point of view when annotating or reviewing an image or video). Don't forget to remove any unnecessary filters before diving back into work on a dataset!
Once a class or a tag has been added to your dataset, it will appear in the list below the workflow filters.
You can type to filter, and filter for multiple classes/tags at once. This will reveal all images or videos that contain at least one of the classes and tags that have been filtered for.
Use the arrows to sort by number of instances. By default, classes and tags will be sorted alphabetically. Click X to clear your filters
You can sort all of the data in your dataset (or a subset that you've filtered for) in the top right of the data page. Your data will be sorted by Date added by default, but you can change this to Date modified, Image dimensions, File size, Filename, and Priority by ascending or descending order.
If your filtering and sorting has revealed a subset of images or videos that you'd like to store separately within your dataset, you can create a new folder by selecting multiple images, and creating a new folder.
Filters are applied on a folder level - the filters that you've applied in one folder will not be carried over to the full dataset.
Updated almost 2 years ago