Create and use tags

Tags, or Classification Tags can be used to tag data for training classifiers, or simply to more effectively filter for related images and videos. Tags, unlike other annotation types like Polygons, are applied to the whole image. They do not have coordinates, but they may have a start-frame and duration when applied to videos or sequences.

To add a new tag, open up an individual image or video from one of your datasets, and start typing in the Tags field.  Click Create at the bottom of the page, or hit enter and your new tag will be saved.


Now that you’ve got a saved tag, you can add it to any of the images and videos in your datasets by selecting it from the Tags menu


You can also do this in bulk. In the Data page of any of your datasets, select the items that you would like to tag.  Create a new tag, or highlight an existing one, and select Apply tag. To remove a tag, click the X to the right.


Tip: The tag applier in the bottom right has a few other handy controls. X allows you to clear all filters, while the left arrow and right arrow sort by alphabetic order, and number of instances of each tag and class respectively

222222 231231

Tagging video

When adding a tag to a video, V7 will automatically apply the tag to the first 30 frames of the video.  You can elongate or shorten the number of frames that a tag covers by dragging its extremities, or the whole block to move it to the relevant section of the video in the timeline at the bottom of the page.


Copying tags

When you add a tags to a dataset, it will initially only appear in the dataset in which it was created.  You can easily copy the tags and classes you’ve created to another dataset in the Classes page.

Select the three dots to the top right of any tag, and select Copy to Dataset.


Filtering & Exporting

Finally, once you've tagged any items within a dataset, you can filter for tagged items in the data tab. You can also export your data using the Export button in the top right of the page. Any tags that have been applied will be reflected in your exported data.