[otb-users] Train Vector Classifier: Shift Scale Sample list Filter error

[Katie Sanders]

I am doing an object based classification. I have completed the 4 steps of
the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.

C:\WINDOWS\system32>SET LC_NUMERIC=C


C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable


C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"


C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"


C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"


C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass"
-classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"

2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB

2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB

2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads

2018-11-08 13:49:24 (INFO): Reading vector file 1/2

2018-11-08 13:49:24 (INFO): Reading vector file 2/2

2018-11-08 13:49:24 (FATAL): itk::ERROR:
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18


Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?


Katie
--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Etienne MAHE]

Hi,

Yes, if you previousely use the module Compute Image Order second
statistics to generate an xml file, the number of features must match to
the number of bands of the image used for the classifications. From my
experience for classification of aerial photos (rgb and nir), only mean and
medians seem revelant. As for example, with six bands, you can use up to
six features.
And just to let you know, the libsvm classifier seems to have issues with
the last otb version when you enable parameters optimization. Awaiting a
bug fix, I use the 6.4.0 which works fine. You should find this version on
OTB repository.

Sincerely,
Etienne.

Post by Katie Sanders
I am doing an object based classification. I have completed the 4 steps of
the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.
C:\WINDOWS\system32>SET LC_NUMERIC=C
C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable
C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"
C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"
C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"
C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass" -
classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"
2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB
2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB
2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads
2018-11-08 13:49:24 (INFO): Reading vector file 1/2
2018-11-08 13:49:24 (INFO): Reading vector file 2/2
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18
Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?
Katie
--
--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Etienne MAHE]

My mistake, it is the compute image second order statistics.

Post by Etienne MAHE
Hi,
Yes, if you previousely use the module Compute Image Order second
statistics to generate an xml file, the number of features must match to
the number of bands of the image used for the classifications. From my
experience for classification of aerial photos (rgb and nir), only mean and
medians seem revelant. As for example, with six bands, you can use up to
six features.
And just to let you know, the libsvm classifier seems to have issues with
the last otb version when you enable parameters optimization. Awaiting a
bug fix, I use the 6.4.0 which works fine. You should find this version on
OTB repository.
Sincerely,
Etienne.

Post by Katie Sanders
I am doing an object based classification. I have completed the 4 steps
of the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.
C:\WINDOWS\system32>SET LC_NUMERIC=C
C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable
C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"
C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"
C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"
C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass" -
classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"
2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB
2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB
2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads
2018-11-08 13:49:24 (INFO): Reading vector file 1/2
2018-11-08 13:49:24 (INFO): Reading vector file 2/2
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18
Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?
Katie
--
--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Katie Sanders]

Thank you! The train vector classifier tool completed fine with the
parameters optimization enabled, is it the trainer or the train vector
classifier that has a bug or the vector classifier where the bug happens?

Post by Etienne MAHE
Hi,
Yes, if you previousely use the module Compute Image Order second
statistics to generate an xml file, the number of features must match to
the number of bands of the image used for the classifications. From my
experience for classification of aerial photos (rgb and nir), only mean and
medians seem revelant. As for example, with six bands, you can use up to
six features.
And just to let you know, the libsvm classifier seems to have issues with
the last otb version when you enable parameters optimization. Awaiting a
bug fix, I use the 6.4.0 which works fine. You should find this version on
OTB repository.
Sincerely,
Etienne.

Post by Katie Sanders
I am doing an object based classification. I have completed the 4 steps
of the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.
C:\WINDOWS\system32>SET LC_NUMERIC=C
C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable
C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"
C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"
C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"
C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass" -
classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"
2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB
2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB
2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads
2018-11-08 13:49:24 (INFO): Reading vector file 1/2
2018-11-08 13:49:24 (INFO): Reading vector file 2/2
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18
Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?
Katie
--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html
You received this message because you are subscribed to the Google
Groups "otb-users" group.
<javascript:>
To unsubscribe from this group, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Etienne MAHE]

It is only for libsvm train vector classifier. When the option parameters
optimization is enabled, the trainer tries to optimize but without stopping
(the process is performed again, and again like a loop without ending with
the confusion matrix), whatever the kernel used (same issue for rbf,
polynomial, sigmoid and linear). I don't know which new code has been
written since 6.6.0 and that could explain this bug. I have also noticed
that this issue appears on both windows and linux systems.

Post by Katie Sanders
Thank you! The train vector classifier tool completed fine with the
parameters optimization enabled, is it the trainer or the train vector
classifier that has a bug or the vector classifier where the bug happens?

Post by Etienne MAHE
Hi,
Yes, if you previousely use the module Compute Image Order second
statistics to generate an xml file, the number of features must match to
the number of bands of the image used for the classifications. From my
experience for classification of aerial photos (rgb and nir), only mean and
medians seem revelant. As for example, with six bands, you can use up to
six features.
And just to let you know, the libsvm classifier seems to have issues with
the last otb version when you enable parameters optimization. Awaiting a
bug fix, I use the 6.4.0 which works fine. You should find this version on
OTB repository.
Sincerely,
Etienne.

Post by Katie Sanders
I am doing an object based classification. I have completed the 4 steps
of the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.
C:\WINDOWS\system32>SET LC_NUMERIC=C
C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable
C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"
C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"
C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"
C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass" -
classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"
2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB
2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB
2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads
2018-11-08 13:49:24 (INFO): Reading vector file 1/2
2018-11-08 13:49:24 (INFO): Reading vector file 2/2
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18
Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?
Katie
--
--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--

--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Etienne MAHE]

Just to finish, there is no issue, neither with other train vector
classifier (such as random forest or knn), nor with vector classifier.

Post by Etienne MAHE
It is only for libsvm train vector classifier. When the option parameters
optimization is enabled, the trainer tries to optimize but without stopping
(the process is performed again, and again like a loop without ending with
the confusion matrix), whatever the kernel used (same issue for rbf,
polynomial, sigmoid and linear). I don't know which new code has been
written since 6.6.0 and that could explain this bug. I have also noticed
that this issue appears on both windows and linux systems.

Post by Katie Sanders
Thank you! The train vector classifier tool completed fine with the
parameters optimization enabled, is it the trainer or the train vector
classifier that has a bug or the vector classifier where the bug happens?

Post by Etienne MAHE
Hi,
Yes, if you previousely use the module Compute Image Order second
statistics to generate an xml file, the number of features must match to
the number of bands of the image used for the classifications. From my
experience for classification of aerial photos (rgb and nir), only mean and
medians seem revelant. As for example, with six bands, you can use up to
six features.
And just to let you know, the libsvm classifier seems to have issues
with the last otb version when you enable parameters optimization. Awaiting
a bug fix, I use the 6.4.0 which works fine. You should find this version
on OTB repository.
Sincerely,
Etienne.

Post by Katie Sanders
I am doing an object based classification. I have completed the 4 steps
of the Large Scale Mean Shift Segmentation in OTB. I am trying to use the
TrainVectorClassifier tool but am getting an error regarding the Field
names for training features. The training shapefile has means and variances
of 18 bands that I am trying to use as training features. I am using OTB
6.6 with QGIS 3. Here is the log of the application.
C:\WINDOWS\system32>SET LC_NUMERIC=C
C:\WINDOWS\system32>SET GDAL_DRIVER_PATH=disable
C:\WINDOWS\system32>SET "GDAL_DATA=C:\OTB-6.6.0-Win64\share\data"
C:\WINDOWS\system32>SET "GEOTIFF_CSV=C:\OTB-6.6.0-Win64\share\epsg_csv"
C:\WINDOWS\system32>SET
"OTB_APPLICATION_PATH=C:\OTB-6.6.0-Win64\lib\otb\applications"
C:\WINDOWS\system32>"C:\OTB-6.6.0-Win64\bin\otbApplicationLauncherCommandLine.exe"
TrainVectorClassifier C:/OTB-6.6.0-Win64/lib/otb/applications -cfield
"class" -classifier "libsvm" -classifier.libsvm.c "100000.0"
-classifier.libsvm.k "rbf" -classifier.libsvm.m "oneclass" -
classifier.libsvm.nu "0.001" -feat "meanB0" "meanB1" "meanB2" "meanB3"
"meanB4" "meanB5" "meanB6" "meanB7" "meanB8" "meanB9" "meanB10" "meanB11"
"meanB12" "meanB13" "meanB14" "meanB15" "meanB16" "meanB17" "varB0" "varB1"
"varB2" "varB3" "varB4" "varB5" "varB6" "varB7" "varB8" "varB9" "varB10"
"varB11" "varB12" "varB13" "varB14" "varB15" "varB16" "varB17"
-io.confmatout "B:/Masters Project/Wowetta/Classification/Confusion
Matrix/svmmatrix1.csv" -io.out "B:/Masters
Project/Wowetta/Classification/Model/svmmodel1.txt" -io.stats "B:/Masters
Project/Wowetta/Classification/p_imagestatistics.xml" -io.vd "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile2_2_training.shp" -v "True"
-valid.vd "B:\Masters
Project\Wowetta\Classification\Training\w_Tile1_1training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_1_training.shp" "B:\Masters
Project\Wowetta\Classification\Training\w_tile3_2_training.shp"
2018-11-08 13:49:24 (INFO): Default RAM limit for OTB is 128 MB
2018-11-08 13:49:24 (INFO): GDAL maximum cache size is 1634 MB
2018-11-08 13:49:24 (INFO): OTB will use at most 8 threads
2018-11-08 13:49:24 (INFO): Reading vector file 1/2
2018-11-08 13:49:24 (INFO): Reading vector file 2/2
ShiftScaleSampleListFilter(00000239450C43A0): Inconsistent measurement
vector size : Input Sample List size 36 Scale measurement vector size 18
Shift measurement vector size 18
Can Anyone help me with this error and how to remedy it? I understand I
have 18 mean features and 18 variance features. Which together is the
sample list of 36. Am I only able to use 18 of the features?
Katie
--
--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--

--
Check the OTB FAQ at
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, send email to
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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
For more options, visit https://groups.google.com/d/optout.

--
--
Check the OTB FAQ at
http://www.orfeo-toolbox.org/FAQ.html

You received this message because you are subscribed to the Google
Groups "otb-users" group.
To post to this group, send email to otb-***@googlegroups.com
To unsubscribe from this group, send email to
otb-users+***@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/otb-users?hl=en
---
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 otb-users+***@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.