HTML
Guide for labeling HTML data.
Overview
The HTML editor is a powerful way for you to annotate data that must be visualized in a specific manner for your annotation workforce. When you are annotating an HTML file, Labelbox will render the HTML page in the pane where our editor would normally render (seen below).
The HTML editor currently only supports classification-type tasks (radio, checklist, and free text).
Some common use cases for the HTML editor are:
- Comparing two objects
- Doing ranking tasks on multiple assets
- Classifying text that must be formatted in a specific manner
- Annotating public webpages that have been saved as HTML files
For information on importing HTML files to Labelbox, see our docs on HTML
Supported annotation types
Below are the annotation types that you may include in your ontology for labeling image data. Classification-type annotations can be applied at the global level and/or nested within an object-type annotation.
| Import annotations | Export annotations | |
|---|---|---|
| Radio classification | Not supported | See payload |
| Checklist classification | Not supported | See payload |
| Free-form text classification | Not supported | See payload |
Sample Use Cases
Comparison of different products
One common use case among our customers is to do a comparison task between two objects or products. This is especially important for any team-building algorithms to rank or compare similar objects.
In order to visualize the two products and all the characteristics of the products, we were able to create a custom HTML page that renders all the information needed to help the annotation team add the necessary classification annotations
We created a custom HTML that can pull in information to compare products. For more details about this, please reach out to [email protected]
Annotating public websites
We do not support direct linking to webpages via their public URL
Most websites do not support other webpages or applications to open their pages in iframes as a security best practice. As a result, if you do try to link public websites (such as https://google.com or https://yahoo.com) they will NOT render in the HTML panel even if they are html pages.
In order to annotate websites, you will need to first download the webpages of interest as HTML files. This can be done programmatically or manually as seen below:
Navigate to a page and right-click to save it as an HTML page.
Once you have the website saved as an HTML file, you can either directly upload it to Labelbox or store it in your cloud storage and send Labelbox a URL link to the HTML. After the HTML page has been uploaded to a dataset in Labelbox, you can queue it for labeling and see that the website is rendered in the labeling flow:
The website is loaded into Labelbox as an HTML file.
Updated 2 months ago