I am conducting image segmentation using otbcli_LargeScaleMeanShift on a series of Sentinel-1 time series images (30 images) that have been preprocessed. My goal is to use this segmentation to apply a Random Forest model for classifying farm fields.
Configuration setup
My system: Processor Intel(R) Xeon(R) Gold 6230 CPU @ 2.10GHz 2.10 GHz (2 processors)
Installed RAM 128 GB
System type 64-bit operating system, x64-based processor
Version of the OTB: 9.1.1
I am Using OTB in the Anaconda Prompt
QGIS version: 3.40.8
Python version: 3.13
Description of my issue
I am conducting image segmentation using otbcli_LargeScaleMeanShift on a series of Sentinel-1 time series images (30 images) that have been preprocessed. My goal is to use this segmentation to apply a Random Forest model for classifying farm fields.
The segmentation process was successful in these two scenarios:
- It generated image segments when I applied the algorithm to 16 out of the 30 images for a region measuring 14,000 km².
- It also produced image segments when I applied the algorithm to all 30 images for a smaller area of interest (AOI). (top right image)
The segmentation process was not successful in the third scenario:
- When I applied segmentation to all the images for a region of 14,000 km², the resulting image segments were incomplete, with some parts missing and others appearing squeezed. I’m unsure why this issue is occurring. (Left image and bottom right image)
Parameter used:
(
f"otbcli_LargeScaleMeanShift -in {ras} -spatialr 20 -ranger 6 -minsize 50 "
f"-tilesizex 8192 -tilesizey 8192 -mode vector -mode.vector.out {seg_shp} "
f"-cleanup false -ram 100128", 1, ‘Image segmentation’
)
Could you help me resolving this error? I tried running twice which still gives me the same output.
Thanks,
Mona