Commit 804001ae authored by Cédric Traizet's avatar Cédric Traizet

Merge branch 'imageRegression' into 'develop'

Image regression application

See merge request !587
parents 6f477490 ce56301f
Pipeline #2554 passed with stages
in 121 minutes and 53 seconds
......@@ -853,17 +853,6 @@ element is equal to 3 pixels, which corresponds to a ball included in a
7 x 7 pixels square. Pixels with more than one majority class keep their
original labels.
The machine learning models in OpenCV and LibSVM also support a
regression mode: they can be used to predict a numeric value (i.e. not
a class index) from an input predictor. The workflow is the same as
classification. First, the regression model is trained, then it can be
used to predict output values. The applications to do that are and .
.. |image_61| image:: ../Art/MonteverdiImages/classification_chain_inputimage.jpg
.. |image_62| image:: ../Art/MonteverdiImages/classification_chain_fancyclassif_CMR_input.png
.. |image_63| image:: ../Art/MonteverdiImages/classification_chain_fancyclassif_CMR_3.png
......@@ -877,144 +866,77 @@ used to predict output values. The applications to do that are and .
Figure 6: From left to right: Original image, fancy colored classified image and regularized classification map with radius equal to 3 pixels.
The input data set for training must have the following structure:
- *n* components for the input predictors
- one component for the corresponding output value
The application supports 2 input formats:
- An image list: each image should have components matching the
structure detailed earlier (*n* feature components + 1 output value)
- A CSV file: the first *n* columns are the feature components and the
last one is the output value
If you have separate images for predictors and output values, you can
use the application.
otbcli_ConcatenateImages -il features.tif output_value.tif
-out training_set.tif
Statistics estimation
The machine learning models in OpenCV, LibSVM and SharkML also support a
regression mode: they can be used to predict a numeric value (i.e. not
a class index) from an input predictor. The workflow is the same as
classification. First, the regression model is trained, then it can be
used to predict output values.
As in classification, a statistics estimation step can be performed
before training. It allows to normalize the dynamic of the input
predictors to a standard one: zero mean, unit standard deviation. The
main difference with the classification case is that with regression,
the dynamic of output values can also be reduced.
Two applications are available for training:
The statistics file format is identical to the output file from
application, for instance:
- `TrainVectorRegression` can be used to train a classifier with a set of geometries
containing a list of features (predictors) and the corresponding output value:
<?xml version="1.0" ?>
<Statistic name="mean">
<StatisticVector value="198.796" />
<StatisticVector value="283.117" />
<StatisticVector value="169.878" />
<StatisticVector value="376.514" />
<Statistic name="stddev">
<StatisticVector value="22.6234" />
<StatisticVector value="41.4086" />
<StatisticVector value="40.6766" />
<StatisticVector value="110.956" />
In the application, normalization of input predictors and output values
is optional. There are 3 options:
- No statistic file: normalization disabled
- Statistic file with *n* components: normalization enabled for input
predictors only
- Statistic file with *n+1* components: normalization enabled for
input predictors and output values
If you use an image list as training set, you can run application. It
will produce a statistics file suitable for input and output
normalization (third option).
otbcli_TrainVectorRegression -io.vd samples.sqlite
-cfield predicted
-io.out model.rf
-classifier rf
-feat perimeter area width
The validation set `io.valid` is used to compute the mean square error between the original output value and the value
predicted by the computed model. If no validation set is provided the input training sample is used to compute the mean square error.
otbcli_ComputeImagesStatistics -il training_set.tif
-out stats.xml
Initially, the machine learning models in OTB only used classification.
But since they come from external libraries (OpenCV and LibSVM), the
regression mode was already implemented in these external libraries. So
the integration of these models in OTB has been improved in order to
allow the usage of regression mode. As a consequence , the machine
learning models have nearly the same set of parameters for
classification and regression mode.
- `TrainImagesRegression` can be used to train a classifier from multiple pairs of predictor images and label images.
There are two ways to use this application:
.. |image11| image:: ../Art/MonteverdiImages/classification_chain_inputimage.jpg
.. |image12| image:: ../Art/MonteverdiImages/QB_1_ortho_MV_C123456_CM.png
.. |image13| image:: ../Art/MonteverdiImages/classification_chain_inputimage.jpg
.. |image14| image:: ../Art/MonteverdiImages/QB_1_ortho_DS_V_P_C123456_CM.png
It is possible to provide for each input image a vector data file with geometries
corresponding to the input locations that will be used for training. This is achieved by using the `io.vd` parameter.
The `sample.nt` and `sample.nv` can be used to specify the number of sample extracted from the images, for training and
validation, respectively.
.. |image15| image:: ../Art/MonteverdiImages/classification_chain_inputimage.jpg
:scale: 88%
- Decision Trees
otbcli_TrainImagesRegression inputPredictorImage.tif
-io.ip inputLabelImage.tif
-io.vd trainingData.shp
-classifier rf
-io.out model.txt
-sample.nt 1000
-sample.nv 500
- Gradient Boosted Trees
- Neural Network
- Random Forests
Alternatively, if no input vector data is provided, the training samples will be
extracted from the full image extent.
- K-Nearest Neighbors
Two applications are also available for predictions:
The behavior of application is very similar to . From the input data
set, a portion of the samples is used for training, whereas the other
part is used for validation. The user may also set the model to train
and its parameters. Once the training is done, the model is stored in an
output file.
- `VectorRegression` uses a regression machine learning model to predict output values based on a
list of features:
otbcli_TrainRegression training_set.tif
-io.imstat stats.xml
-io.out model.txt
-sample.vtr 0.5
-classifier knn
-classifier.knn.k 5
-classifier.knn.rule median
Once the model is trained, it can be used in application to perform the
prediction on an entire image containing input predictors (i.e. an image
with only *n* feature components). If the model was trained with
normalization, the same statistic file must be used for prediction. The
behavior of with respect to statistic file is identical to:
- no statistic file: normalization off
- *n* components: input only
- *n+1* components: input and output
-in input_vector_data.shp
-feat perimeter area width
-model model.txt
-out predicted_vector_data.shp
The model to use is read from file (the one produced during training).
- Similarly, `ImageRegression` takes an image of predictors as input and computes the output image using a regression model:
otbcli_PredictRegression -in features_bis.tif
-model model.txt
-imstat stats.xml
-out prediction.tif
-in input_image.tif
-model model.txt
-out predicted_image.tif
......@@ -71,8 +71,8 @@ otb_create_application(
NAME PredictRegression
SOURCES otbPredictRegression.cxx
NAME ImageRegression
SOURCES otbImageRegression.cxx
......@@ -84,7 +84,7 @@ private:
SetDocLimitations("The input image must have the same type, order and number of bands as the images used to produce the statistics file and the SVM model file. If a statistics file was used during training by the TrainImagesClassifier, it is mandatory to use the same statistics file for classification. If an input mask is used, its size must match the input image size.");
SetDocSeeAlso("TrainImagesClassifier, ValidateImagesClassifier, ComputeImagesStatistics");
SetDocSeeAlso("TrainImagesClassifier, ComputeImagesStatistics");
......@@ -785,16 +785,16 @@ if(OTB_USE_OPENCV)
#----------- PredictRegression TESTS ----------------
otb_test_application(NAME apTvClPredictRegressionTest_monovar
APP PredictRegression
#----------- ImageRegression TESTS ----------------
otb_test_application(NAME apTvClImageRegressionTest_monovar
APP ImageRegression
OPTIONS -in ${INPUTDATA}/QB_Toulouse_Ortho_PAN.tif
-model ${OTBAPP_BASELINE_FILES}/apTvClTrainRegressionTest_monovar.rf
-imstat ${INPUTDATA}/QB_Toulouse_Ortho_regression.xml
-out ${TEMP}/apTvClPredictRegressionTest_monovar.tif
-out ${TEMP}/apTvClImageRegressionTest_monovar.tif
VALID --compare-image 1
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment