Data ingestion is made easy on Kili Technology.
Supported file formats
For TEXT projects, you can import
.txt files and with a small script, any xml file.
For PDF projects, you can import
Native support Here are the natively supported formats for image :
.dcm). To import DICOM images you can use this recipe.
The tiff format (
.tiff) is not natively supported, we recommend to convert files to png format. Imagemagick can help you do that.
For other custom or specific formats, you can try to add them to a sample project. If the image doesn't load, it means you need to convert it using
For native video labeling, all formats should be readable
For frame labeling (videos are split into images), here are the tested formats :
Native support Most formats should be natively usable. Among them (tested):
All web formats should work. The following formats have been tested (you can display both audio and video files)
From where can you ingest data
Assets can be uploaded from
- your local workstation
- from a public cloud by uploading a CSV file containing the URLs redirecting to each asset
- or from your on-premise servers, without the data being stored on our servers. The data is streamed through the web browser, and only available within your internal network.
How to upload assets
Through the user interface
From the Dataset tab, you click on
Add new. You either
Upload Local Data: drag & drop asset files (one asset / file, 500 files / batch)
Connect Cloud Data: drap & drop a CSV file with URLs to the asset (one asset / line)
You can also import data directly from your Kili datasets.
Through the API
Through the API follow examples on our Github Kili-Playground.
The maximum number of assets per project is limited to 25000. If you require more you can contact us at firstname.lastname@example.org.