Labeling Geospatial assets (Beta π)
Geospatial projects are currently available in Private beta. To request access, please contact our team.
Introduction
The new Geospatial projects bring powerful capabilities tailored for working with satellite and aerial imagery at scale. Designed to support high-performance workflows, these projects enable precise and efficient annotation of complex geospatial datasets.
Key features include:
- Native Cloud-Optimized GeoTIFF (COG) support: Load and visualize large geospatial images instantly with dynamic tiling and on-demand renderingβno need for pre-processing or downsampling.
- Multispectral and Hyperspectral image support: Handle rich imagery beyond RGB. You can create and visualize custom composites (e.g., NDVI, false color) using any combination of bands to enhance feature visibility and annotation quality.
- Larger image and label support: Seamlessly work with high-resolution geospatial assets and detailed annotations, without compromising performance or usability.
- Streamlined navigation on large assets: Quickly zoom in on specific objects or geographic locations to accelerate labeling and improve accuracy across vast image surfaces.
These capabilities allow teams to work directly with the data they need, in the formats they use most, all within a streamlined labeling interface built for geospatial tasks.
Setting up a geospatial project
Create a new project
From the project creation page, select Geospatial as the project type. This ensures your project is configured to support large image assets, multi-band data, and geospatial labeling tools.
Configure your interface, team, and workflow
Add the necessary project members and assign them appropriate roles (e.g., labelers, reviewers, managers). Then, set up your labeling workflow according to your quality and collaboration requirements.
(You can refer to Customize your interface, Managing Project Members and Quality Workflow Setup for more details.)
Once your project is created, you're ready to start importing geospatial assets and labeling them using our dedicated tools.
Adding geospatial assets
To import assets into your Geospatial project, you can follow the standard procedures for uploading or synchronizing assets.
Refer to our Managing Assets documentation for a complete guide.
Below are the specifics to keep in mind when working with geospatial data.
Uploading assets
When uploading assets directly, ensure your files meet the format and size requirements for geospatial workflows.
-
Supported file formats:
Geospatial projects support GeoTIFF-based formats, including.tif
,.jp2
, and.ntf
. -
Recommended file size:
Geospatial projects are not restricted to the 1GB limit that applies to standard image projects. However, for optimal performance, we recommend keeping files under 5GB. -
Supported projection systems: EPSG4326(WGS 84) or EPSG3857 (Web Mercator)
-
Multi-layered assets: You can upload multiple files as individual layers of the same asset, allowing you to work with different processing levels or sensor types within a unified labeling interface. To learn how to upload multi-layered assets, refer to our SDK tutorial.
Note: Larger files are supported thanks to dynamic tiling and streaming, but performance may vary based on connection speed and system resources.
Synchronizing assets from a cloud storage bucket
You can also connect a cloud storage bucket to automatically sync assets into your project.
File format requirement:
To enable dynamic tile loading in the labeling interface, files in your bucket must be in Cloud-Optimized GeoTIFF (COG) format.
COGs allow the viewer to fetch only the visible portions of the image on demand, which is essential for working with large-scale imagery efficiently.
Labeling a geospatial asset
Navigate the image
You can navigate the image using several tools:
- Pan: Click and drag to move around the image.
- Zoom
- Use your mouse scroll wheel or trackpad to zoom in and out.
- Adjust zoom levels directly from the header or by using the
+
and-
keys - Press
Shift
+ 1 to go back to the initial zoom level - Or, hold the
Shift
key and draw a box around the area you want to zoom into for a quick focus.
- Mini-Map
A mini-map in the corner of the interface shows your current viewport. You can drag the viewport box to quickly jump to a different part of the image. You can hide/show it from the display settings.

- Jump to object
From the Job viewer, you can click on a specific annotation or object in the list. The interface will automatically pan and zoom to its exact location on the image

Customizing image display settings
Rescale image bands
Adjust the minimum and maximum values used to render each band. This helps enhance contrast and visibility of specific features.
By default, the viewer uses the 2nd to 98th percentile of pixel values for rescaling. You can view the histogram of each band to guide your adjustments.
Create RGB composites
When working with multispectral or hyperspectral imagery, you can map any available bands to Red, Green, and Blue channels to generate a custom composite.
This is useful for creating true-color or false-color views, depending on your analysis needs.

Create spectral indices
Define custom band math expressions to compute spectral indices such as NDVI, NDWI, or any other relevant formula.
You can then apply a colormap to better highlight the index values and identify patterns or anomalies in the imagery.

Rotate the image
You can use the rotation button to adjust the orientation of geospatial images. With each click, the image rotates by 10 degrees. Alternatively, you can enter the desired rotation angle directly into the input field for more precise adjustments. If you need to reset the rotation to its original orientation, simply double-click the rotation button.
Spatial tools for precision labeling
Find location by coordinates
Quickly locate a specific point in the image by right-clicking and selecting βFind location,β then entering latitude and longitude.

Copying coordinates
You can quickly copy the geo-coordinates of any point. To do that, hover over it, click the right mouse button, and then select Copy GPS Coordinates. The coordinates are copied to the clipboard.
Measuring distances
You can measure distances between any two points. The distance is provided in meters and feet. This is helpful in situations when an objectβs length is a decisive factor for assigning it a specific class.
Annotate the image
Geospatial projects support a wide range of annotation tools to match your labeling needs:
- Bounding Boxes, Polygons, Points, and Lines
- Semantic Segmentation (including SAM2)
- Classification and Transcriptions
Importing/exporting labels
Exporting labels
Supported Export formats include:
- Kili format: A versatile format for integration with Kili workflows and further processing. Note: In exported annotations,
x
stands for longitude andy
for latitude - GeoJSON : Widely used for geospatial data, compatible with various GIS tools and applications.
Both formats can be used in the Kili app and the SDK. For detailed steps on how to export labels using the SDK, refer to the export documentation here.
Importing labels
For detailed steps on how to import labels using the SDK, refer to the import labels tutorial.
Supported formats include:
- Kili format: Re-import full annotations exported from another Kili project or asset. See Kili data format for more details.
- Shapefiles (
.shp
): Check out our tutorial for more details. - GeoJSON: Convert your GeoJSON files into Kili format using our utility functions before importing. Check out our tutorial for a step-by-step guide.
Updated about 13 hours ago