Imagery

Uploading image files

The easiest way to import images into Picterra is to upload GeoTIFF from your computer.

To do so, enter one of your projects and hover over the “Add Imagery” button. You can then select “Upload your files”:

Image Details

The “Image Details” popup contains more information about your uploaded images. It is available from within a Project by choosing the “Info” option in the drop-down menu for a selected image.

The “Image Details” popup contains the following information:

You can do a number of things from that popup:

Download the image as a Cloud Optimized GeoTiff
Use the “Tiles URL” to connect from a GIS software as a TMS data source:
- To connect to QGIS, replace the {y} in the url with {-y}.
Manage the Detection Areas for your image
Check the area / resolution
Use the raster Id via the API or python libraries
Edit the Multispectral bands to be viewed and the min/max ranges.
Assign a tag to your image: this is a beta feature that allows to filter your images via the Public API

Imagery Servers

You can connect Picterra to remote imagery servers through the “Import from imagery server” option in the “Add imagery” menu. This will open a window to manage your remote imagery servers.

Importing from cloud storage / URL

This is a Beta feature, which means it is not covered by our SLAs and could still have some rough edges. We welcome feedback on the feature!

If you have a link to your file, you can also import it directly by copy/pasting the link in Picterra.

Typically, if your file is stored on a cloud storage solution (e.g. Amazon S3, Azure Blob Storage or Google Cloud Storage), you can generate a public URL to your file and then copy/paste this URL in Picterra to import it.

To do so, enter one of your projects, select Add Imagery and the Import from URL option and then enter the URL to your image file as well as a name for it.

After cliking Import, you can monitor the status of your import through the notification center.

Image Streaming Services

Picterra provides Image Streaming Services, allowing you to use our platform as your central image server.

Picterra’s Image Streaming Service capabilities are:

Storage of GeoTIFFs images imported via upload (UI or API) or via public URLs from clouds.
Compatibility with any geospatial sources as rasters GeoTIFFs (e.g. Optical, SAR, hyperspectral, thermal infrared, etc.).
Picterra Image server allows streaming of any stored raster on Picterra served via XYZ / TMS URLs (OGC standard) to external platforms, compatible with all GIS and many CAD softwares (e.g. Spatial Manager for Autocad).
Restreaming of remote image servers (e.g. WMS/TMS) as XYZ / TMS URLs (OGC standard), allowing to centralize various sources in one place.
Image and band selection setup via the “Info” or “Edit multispectral bands” menu.

The “Image Details” popup contains the information about your uploaded images and more specifically the information needed to stream that image to external softwares and platforms.

The “Tiles URL” is providing a TMS URL (OGC standard) to be used to connect from a GIS, CAD or other software as a TMS data source.

Here is an example of such URL: https://app.picterra.ch/tiler/raster/rasters/942ca285e21b4f8e8cc85347430de3e4/normalized.tif/tms/{z}/{x}/{y}
To connect to QGIS as a XYZ data source, replace the {y} in the url with {-y}. https://app.picterra.ch/tiler/raster/rasters/942ca285e21b4f8e8cc85347430de3e4/normalized.tif/tms/{z}/{x}/{-y}
In case of Multispectral rasters, the URL will be slightly different but working the same way.

If you are interested in using the Picterra’s Image Streaming Services, please contact your Customer Success manager or support@picterra.ch. We will provide you with a quote with respect to the amount of data to store and stream per year as well as the estimated usage (# of tiles viewed per month, # users).

Multispectral imagery

Multispectral support allows training detectors on more than 3 bands.

For example, by uploading a 4 bands RGB+NIR image as a multispectral image, a custom detector will use the information from the 4 bands and the user can choose which bands to visualize and how to normalize each band. This choice is done by specifying what we call a bandspec.

Multispectral support allows training detectors on more than 3 bands.

Multispectral imagery support is an advanced platform feature that assumes the user is familiar with geospatial multispectral imagery.

Walkthrough

The choice to upload an image as multispectral happens at upload. You need to check the ‘multispectral’ box in the advanced parameters:

Once uploaded, if you open the raster info popup, you can see the image is indicated as multispectral:

then you can “edit bandspec”

Editing the bandspec

By clicking on edit bandspec, you can edit the bandspec to change the way the multispectral image is interpreted by the platform.

The dialog is made of three tabs.

Visualization

Band ranges

Raw JSON

Visualization

Band ranges

Raw JSON

This a very error-prone way of editing the bandspec; please try to use it only when you want to copy-paste bandspecs from one raster to another.

Here you can edit the “raw” form of the image band specification: you can think of it as if you can modify in parallel the bandspec with the two previous tabs.

The bandspec is a JSON document, so you may want to use an online JSON editor/beautifier to make editing simpler.

The ‘bandspec’ consists of two main elements:

- ranges is an array of arrays specifying, for each band in the image, its minimum and maximum value. This is then used to normalize the channel between 0 and 1 by using (value - min) / (max - min) . This corresponds to the “Band ranges” tab described above.
- display_bands specifies combinations of bands to use for RGB visualization. Each entry has a name (string) and an array of 3 numbers that specify the bands to use for R, G, and B. This corresponds to the “Visualization” tab described above. Example display_bands with 2 entries:

[
  {
    "name": "RGB",
    "bands": [0, 1, 2]
  },
  {
    "name": "DSM",
    "bands": [3, 4, 5]
  }
]

Once you save the bandspec, the way the image is displayed should change when you visualize it.

Custom detector

You can build custom detectors on multispectral images. A detector will always be trained on ALL the bands in the image (not just the ones used in band settings). You also cannot mix images with different number of bands in a detector.

This means that changing the bandspec ranges of an image will impact the detector (because this define how the data is normalized before passing it to the detector). But changing the band settings won’t impact the detector.

Does range matter if I have a Byte/uint8 image ?

No, if your image is a multiband uint8 image, you can keep the default ranges of [0, 255] which is what you see on your screen. You could still change the ranges to e.g. increase contrast though.

ranges is really useful if your image is not uint8. For example if it’s an int16 or float32 image. Then, in order to train a ML model, the image has to be normalized between [0, 255] (or 0-1 if you’re working with floating point) and this is why we need to know the min and max of each band.

Troubleshooting

North up projection error

If you are getting an error when uploading an image because it is not north up, you can fix that by reprojecting the image into a north-up projection like EPSG:4326 or one of the UTM projection.

You can do so in QGIS by selecting the “Warp (Reproject)” menu item and then choosing a suitable Target CRS:

How to Georeference an image?

If your imagery is not already a georeferenced orthomosaic or image, you can georeference it manually using a GIS software and reference points from either another georeferenced raster or ground control points.

Here you will find documentation on how to georeference your image with QGIS using its georeferencing plugin.

I have an error related to ‘WebGL’

WebGL is a technology that allows us to display large number of annotations on the platform.

If you have an error related to WebGL, try to go to http://webglreport.com/?v=2 . You should see a green bar at the top saying that your browser supports WebGL.

If your browser supports WebGL but you still get an error on the platform, try restarting your computer.

If you browser doesn’t support WebGL, you might need to update it to the latest version or update your graphics drivers.

If the error persists, send us an email at support@picterra.ch

Imagery

Uploading image files

Image Details

Imagery Servers

Importing from cloud storage / URL

Image Streaming Services

Multispectral imagery

Walkthrough

Editing the bandspec

Visualization

Band ranges

Raw JSON

Visualization

Band ranges

Here you can reduce the values of the ranges interpreted by the platform. The band ranges will be cut outside them.

Raw JSON

This a very error-prone way of editing the bandspec; please try to use it only when you want to copy-paste bandspecs from one raster to another.

Custom detector

Does range matter if I have a Byte/uint8 image ?

Troubleshooting