Hello,
First of all note that we now have a forum and discussions should occur
there : https://forum.orfeo-toolbox.org/
You can increase the default RAM limit using the "OTB_MAX_RAM_HINT"
environment variable (in MB). In the case of the Segmentation
application you can also set manually the tile size with the
mode.vector.tilesize parameter.
If you are OK with loading the full image in memory you can use the
raster mode instead of the vector mode (-mode raster and
-mode.raster.out rasterFilename). This will produce a raster instead of
a vector and should save you the vectorization computation time.
"Stitching" can be disabled if you don't think it is worth the
computation time (-mode.vector.stitch)
MinSize is an optional step (it will be disabled if you set it to 0). It
does the same thing as the LSMSSmallRegionsMerging, It merges small
segments to bigger segments, I refer you to the associated
documentation,
https://www.orfeo-toolbox.org/packages/nightly/latest/CookBook-6.7/Applications/app_LSMSSmallRegionsMerging.html
.
Hope that helps,
Cédric
|
|
On 25/03/2019 16:16, [email protected] wrote:
Hi,
Segmentation is a long process, but what takes the most time is the
"stitching" phase after segmentation inside the blocks (cf 121 blocks
in the following output).
I have a lot of memory and a lot of cores on my machines. I would like
to run on bigger blocks than those of 1024x1024 pixels (ideally no
blocking -> all in memory) to reduce the stitching phase duration (and
speed-up the all process)
Is it possible ? Is it also possible to increase the tiny 128 MB
Default RAM limit ?
I have another question by the way: what exactly does this parameter?
-filter.meanshift.minsize (At first glance, I believed it was to
control the blocks size but obviously not...)
numactl --cpunodebind=1 --membind=1 otbcli_Segmentation -in
attributes_multiband.tif -mode vector -mode.vector.out segments.shp
-filter meanshift -filter.meanshift.spatialr 20
-filter.meanshift.ranger 20 -filter.meanshift.minsize 10000
2019-03-25 16:04:14 (INFO): Default RAM limit for OTB is 128 MB
2019-03-25 16:04:14 (INFO): GDAL maximum cache size is 4757 MB
2019-03-25 16:04:14 (INFO): OTB will use at most 24 threads
2019-03-25 16:04:14 (INFO): No kwl metadata found in file
/b/home/ipgs/a2s_prod/ON_DEMAND/USERS/David/Urban/JOB_2019-03-22_11-07-55_37
8789/Tiles/31TEM/attributes_multiband.tif
2019-03-25 16:04:14 (INFO): Use threaded Mean-shift segmentation.
2019-03-25 16:04:14 (INFO): Large scale segmentation mode which output
vector data
2019-03-25 16:04:14 (INFO): Estimation will be performed in 121 blocks
of 1024x1024 pixels
Computing meanshift segmentation: 2% [*
]
Best regards,
David
--
This mailing will be abandoned soon in favor of a more friendly forum:
https://forum.orfeo-toolbox.org
Thanks to visit it for any question related to Orfeo Toolbox (OTB)
usage, or to check the OTB FAQ:
http://www.orfeo-toolbox.org/FAQ.html
---
You received this message because you are subscribed to the Google
Groups "otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send
an email to [email protected]
<mailto:[email protected]>.
For more options, visit https://groups.google.com/d/optout.
--
This mailing will be abandoned soon in favor of a more friendly forum:
https://forum.orfeo-toolbox.org
Thanks to visit it for any question related to Orfeo Toolbox (OTB) usage, or to
check the OTB FAQ:
http://www.orfeo-toolbox.org/FAQ.html
---
You received this message because you are subscribed to the Google Groups "otb-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
