Hi Guillaume,
Thanks for the tips on object-based classification. I am in the same
position as Geoffrey and Patricia. I have been able to successfully segment
an image. However I am really struggling with the classification part.
On my most recent attempt after reading your suggestions I tried running
OGRLayerClassifier. When I click Run, the window closes and nothing seems
to have happened.
I am probably missing something obvious.
I’m not sure I understood your suggestion about ‘*Adding several numeric
fields to store the features attached to geometries. **The fields can be
extracted from a raster with LSMSVectorization and modified with QGis.’*
Won’t the training layer just need to have a ClassID field added? The mean
and variance of the bands are the only other required fields? Thanks.
Best Regards,
Mark
Le mardi 3 janvier 2017 21:43:57 UTC+8, Guillaume Pasero a écrit :
>
> Hi,
>
> You should use Train*Vector*Classifier instead of TrainImagesClassifier.
> This will train and produce a SVM model ("-io.out" output file). You will
> have to set (at least) :
>
> - Input Vector Data (io.vd)
> - Output model (io.out)
> - Field names for training features (feat)
> - Field containing the class id for supervision (cfield, in your case
> it will be "training")
> - Classifier to use for the training (classifier = libsvm)
>
> However, before using this application, you will have to prepare a vector
> dataset with only your training polygons (polygons with a missing value in
> the field "training" will make the application crash).
>
> Once you have the output model file (simple text/xml file) you can use it
> to do the classification on the full dataset (using application
> OGRLayerClassification).
>
> Regards,
> Guillaume
>
> On 01/02/2017 11:54 PM, Geoffrey Balme wrote:
>
> I forgot to say that after the segmentation, I added a field in my vector
> attribute table named "training", and I gave some numeric values to few
> polygons based on the classes I want the segments to be classified in (3
> classes)
>
> So now I'm with a few polygons having a training value, and most of the
> polygon having no training value, and I want to do the svm classification
> from this.
>
> Le lundi 2 janvier 2017 23:52:47 UTC+1, Geoffrey Balme a écrit :
>>
>> Hi Guillaume,
>>
>> I'm also trying to do supervised object-based classification.
>> What I did so far is a meanshift segmentation on my satellite raster
>> image.
>> Then I did TrainImagesClassifier (svm) on the vector results of the
>> segmentation.
>> And at this point I'm not sure anymore what to do as there was no output
>> on the previous step.
>> Could you guide me from the segmentation to have an object-based
>> classification using SVM ?
>>
>> (I tried to do OGRLayerClassification, but had no idea of what to put in
>> the xml inputs etc..., the only input I have are my raster and my
>> segmentation vector result)
>>
>> Thank you !
>>
>> Le jeudi 24 novembre 2016 14:46:00 UTC+1, Guillaume Pasero a écrit :
>>>
>>> Hi Patricia,
>>>
>>> If I understand right, you want to perform object-based classification.
>>>
>>> In OTB-5.6.1 there is the application TrainVectorClassifier to train a
>>> classifier based on input geometries. You have to set several fields in the
>>> geometries you want to use for training :
>>>
>>> - one field to indicate the class of the geometry
>>>
>>> - several numeric fields to store the features attached to geometries.
>>>
>>> The fields can be extracted from a raster with LSMSVectorization and
>>> modified with QGis.
>>>
>>> After the training, you can either apply the model on the full set of
>>> geometries (with OGRLayerClassifier), or try to apply it on a raster (with
>>> ImageClassifier, but more difficult).
>>>
>>> Regards,
>>>
>>> Guillaume
>>> On 11/24/2016 02:09 PM, Patricia Lourenco wrote:
>>>
>>> Dear all,
>>> I am new with OTB-Monteverdi 5-6-1.
>>>
>>> I want to classify an image based on the segments created in the
>>> LSMVSVectorization (step 4 of the segmentation)using OTB/Monteverdi
>>> versions 5.6.1.
>>> However, I am not being able to do it.
>>>
>>> My questions are:
>>>
>>> 1. Which are the steps that I should take to do a classification based
>>> on segmentation?
>>>
>>> 2. Which OTB-Application should I use to select the segments for my
>>> classes?
>>>
>>> Thank you, in advance, for your help.
>>>
>>> Sincerely,
>>> Patricia
>>> --
>>> --
>>> 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 [email protected]
>>> To unsubscribe from this group, send email to
>>> [email protected]
>>> 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 [email protected].
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>>
>>> --
>>> <http://www.c-s.fr> *Guillaume PASERO*
>>> Responsable technique
>>> *Business Unit E-SPACE & Geo Information - Département Image &
>>> Applications*
>>>
>>> *CS Systèmes d'Information*
>>> Parc de la Grande Plaine - 5, Rue Brindejonc des Moulinais - BP 15872
>>> 31506 Toulouse Cedex 05 - FRANCE
>>> +33 561 17 64 21 - [email protected]
>>>
>> --
> --
> 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 [email protected]
> <javascript:>
> To unsubscribe from this group, send email to
> [email protected] <javascript:>
> 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 [email protected] <javascript:>.
> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> <http://www.c-s.fr> *Guillaume PASERO*
> Responsable technique
> *Business Unit ESPACE & GeoInformation - Département Payload Data &
> Applications*
>
> *CS Systèmes d'Information*
> Parc de la Grande Plaine - 5, Rue Brindejonc des Moulinais - BP 15872
> 31506 Toulouse Cedex 05 - FRANCE
> +33 561 17 64 21 - [email protected] <javascript:>
>
--
--
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 [email protected]
To unsubscribe from this group, send email to
[email protected]
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 [email protected].
For more options, visit https://groups.google.com/d/optout.
