Segmentation not producing any output

Dear all,

I wondering if any of you can explain why I am not getting any entries in the shape file produced by the OTB segmentation algorithm. I have downloaded a set of Sentinel 2LA from the European Space Agency. Converted them for spectral reflectance using the SCP Semi Automated Classifer plugin and clipped them after conversion. The output of the algorithm is as follows:

The input file is the virtual band set - that was produced in SCP Semi-Automation Classification.

I am running the following Plugins and versions of QGIS:

QGIS version
3.32.0-Lima
QGIS code revision
311a8cb8a6
Qt version
5.15.3
Python version
3.9.5
GDAL/OGR version
3.7.0
PROJ version
9.2.1
EPSG Registry database version
v10.088 (2023-05-13)
GEOS version
3.11.2-CAPI-1.17.2
SQLite version
3.41.1
PDAL version
2.5.3
PostgreSQL client version
15.2
SpatiaLite version
5.0.1
QWT version
6.1.6
QScintilla2 version
2.13.1
OS version
Windows 10 Version 2009

Active Python plugins
opeNoise
2.0
pca4cd
23.2a
pluginbuilder3
3.2.1
SemiAutomaticClassificationPlugin
7.10.11
TomBio
3.4.2
ViewshedAnalysis
1.8
db_manager
0.1.20
MetaSearch
0.3.6
otbprovider
2.12.99
processing
2.12.99

I hope that someone can shed some light on what is happening and how to resolve.

Regards,
Alasdair

The attribute table of the Shapefile that is produced by the segmentation algorithm is as follows:

It shouldn’t be empty - is there any reason as to why it is empty? And how to resolve that error.

Hi Grant,

It seems that there are zero values in your Sentinel image stack. Have you checked that the image has the correct values and that it’s not made up entirely of zero values?
There are several things you can try:

  • Check your vrt on Qgis by checking its values
  • Use the GDAL tool ‘fill null values (nodata)’ to improve your image
  • Try segmentation only on an S2 L2A tiff image to see if it’s the vrt that’s blocking or the image itself.

It may be the process you have carried out on SCP that makes the image uninterpretable for the tool.

Kinds regards,
Adrien