Import using URLs
Shows how to import data from public or presigned URLs into a Labelbox dataset.
You can use the Python SDK to import individual files using public or presigned URLs. You can upload up to 150,000 individual files in a single operation.
To do so, you need an Labelbox API key and a Python script that lists the URLs you import.
The Labelbox provides convenient tools to simplify the process, including sample scripts customized for the following data types:
- Image
- Video
- Text
- Tiled imagery
- Documents (PDFs)
- Chat (conversational text)
- Audio
- HTML
Create import script
To import data using public or pre-signed URLs, you need to create a Python script that includes the names of the files to import. You can use the Labelbox app to handle the initial work:
- From the default dataset screen, select Import data from cloud buckets securely.
- When the Connection type prompt appears, select Use public URLs, upload files, or import raw text.
- Create a new API key by entering a name and selecting the Generate button.
When you do this, your new API key is added to the script and copied to the Clipboard. - Save a copy of your new API key in a safe location. You can paste from the Clipboard into a secret store. Strictly speaking, this is optional; however, we strongly recommend saving the key before the Clipboard copy is lost or replaced. (You cannot recover lost keys.)
- Save and customize the sample Python script for your data.
To learn more about the script and your available options, see:
- Use Python SDK
- More about specific data types, including:
- Image
- Video
- Text
- Geospatial
- Documents (PDFs)
- Conversational text (chat)
- Audio
- HTML
Updated about 2 months ago