add_sample_identifier_training_dataT_add_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data (Operator)

Name

add_sample_identifier_training_dataT_add_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data — Add training data to an existing sample identifier.

Warning

add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data is obsolete and is only provided for reasons of backward compatibility. New applications should use the operators for Deep-Learning-based classification instead, for details see Deep Learning / Classification.

Signature

add_sample_identifier_training_data(SampleImage : : SampleIdentifier, ObjectIdx, GenParamName, GenParamValue : ObjectSampleIdx)

Description

add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data adds training data to an existing sample identifier. This is a prerequisite for the training of the sample identifier with the operator train_sample_identifiertrain_sample_identifierTrainSampleIdentifierTrainSampleIdentifiertrain_sample_identifier.

For an explanation of the concept of sample-based identification see the introduction of chapter Legacy / Identification.

In order to be able to identify objects, the SampleIdentifierSampleIdentifierSampleIdentifiersampleIdentifiersample_identifier must be trained with a representative set of sample images of the objects to be identified. These sample images must be added to the SampleIdentifierSampleIdentifierSampleIdentifiersampleIdentifiersample_identifier with add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data.

With each call of add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data, one SampleImageSampleImageSampleImagesampleImagesample_image is added to the SampleIdentifierSampleIdentifierSampleIdentifiersampleIdentifiersample_identifier. The complete training data, which consists of all the sample images that have been added to the sample identifier with multiple calls of add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data, must cover

all objects to be identified and
all different views of the objects that may appear during the identification process, sampled to at least 45°.

The domain of the SampleImageSampleImageSampleImagesampleImagesample_image should be reduced to the object, which is visible in the image. This is necessary to avoid that the SampleIdentifierSampleIdentifierSampleIdentifiersampleIdentifiersample_identifier is trained to the background of the sample images.

Note that in contrast to the preparation data, it is essential that the requirements mentioned above are fulfilled. Otherwise, the sample identifier will not be able to identify the objects for which relevant sample images are missing.

Also note that you must provide RGB color images, if the generic parameter 'add_color_info'"add_color_info""add_color_info""add_color_info""add_color_info" has been set to 'true'"true""true""true""true" in create_sample_identifiercreate_sample_identifierCreateSampleIdentifierCreateSampleIdentifiercreate_sample_identifier.

Ideally, the images used for preparation and training (see add_sample_identifier_preparation_dataadd_sample_identifier_preparation_dataAddSampleIdentifierPreparationDataAddSampleIdentifierPreparationDataadd_sample_identifier_preparation_data and prepare_sample_identifierprepare_sample_identifierPrepareSampleIdentifierPrepareSampleIdentifierprepare_sample_identifier) are identical. See prepare_sample_identifierprepare_sample_identifierPrepareSampleIdentifierPrepareSampleIdentifierprepare_sample_identifier for an explanation on how to reuse preparation data for the training of the sample identifier.

The following generic parameters can be used to influence the behavior of the operator add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data. These parameters and their corresponding values can be specified by using GenParamNameGenParamNameGenParamNamegenParamNamegen_param_name and GenParamValueGenParamValueGenParamValuegenParamValuegen_param_value, respectively. The following values for GenParamNameGenParamNameGenParamNamegenParamNamegen_param_name are possible:

'image_resize_method'"image_resize_method""image_resize_method""image_resize_method""image_resize_method":

See create_sample_identifiercreate_sample_identifierCreateSampleIdentifierCreateSampleIdentifiercreate_sample_identifier for a description of this parameter.

List of values: 'none'"none""none""none""none", 'scale_factor'"scale_factor""scale_factor""scale_factor""scale_factor", 'subsampling_step'"subsampling_step""subsampling_step""subsampling_step""subsampling_step", 'image_area'"image_area""image_area""image_area""image_area"

Default value: If the 'image_resize_method'"image_resize_method""image_resize_method""image_resize_method""image_resize_method" is not set explicitly by this operator, the value that has been set with create_sample_identifiercreate_sample_identifierCreateSampleIdentifierCreateSampleIdentifiercreate_sample_identifier or set_sample_identifier_paramset_sample_identifier_paramSetSampleIdentifierParamSetSampleIdentifierParamset_sample_identifier_param will be used.

'image_resize_value'"image_resize_value""image_resize_value""image_resize_value""image_resize_value":

See create_sample_identifiercreate_sample_identifierCreateSampleIdentifierCreateSampleIdentifiercreate_sample_identifier for a description of this parameter.

Suggested values: 0.25, 0.5, 1.0, 2, 3, 4

Default value: If the 'image_resize_value'"image_resize_value""image_resize_value""image_resize_value""image_resize_value" is not set explicitly by this operator, the value that has been set with create_sample_identifiercreate_sample_identifierCreateSampleIdentifierCreateSampleIdentifiercreate_sample_identifier or set_sample_identifier_paramset_sample_identifier_paramSetSampleIdentifierParamSetSampleIdentifierParamset_sample_identifier_param will be used.

add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data returns the object sample index of the training sample given in SampleImageSampleImageSampleImagesampleImagesample_image. This index can, e.g., be used to remove this sample from the training data, if the sample identifier should be retrained based on a different set of training data.

Execution Information

Multithreading type: reentrant (runs in parallel with non-exclusive operators).
Multithreading scope: global (may be called from any thread).
Automatically parallelized on internal data level.

Parameters

SampleImageSampleImageSampleImagesampleImagesample_image (input_object) (multichannel-)image → object (byte)

Image that shows an object.

SampleIdentifierSampleIdentifierSampleIdentifiersampleIdentifiersample_identifier (input_control) sample_identifier → (handle)

Handle of the sample identifier.

ObjectIdxObjectIdxObjectIdxobjectIdxobject_idx (input_control) integer → (integer / string)

Index of the object visible in the SampleImageSampleImageSampleImagesampleImagesample_image.

Suggested values: 0, 1, 2, 3, 4, 5

GenParamNameGenParamNameGenParamNamegenParamNamegen_param_name (input_control) attribute.name-array → (string)

Generic parameter name.

Default: []

List of values: 'image_resize_method'"image_resize_method""image_resize_method""image_resize_method""image_resize_method", 'image_resize_value'"image_resize_value""image_resize_value""image_resize_value""image_resize_value"

GenParamValueGenParamValueGenParamValuegenParamValuegen_param_value (input_control) attribute.value-array → (real / string / integer)

Generic parameter value.

Default: []

Suggested values: 'none'"none""none""none""none", 'scale_factor'"scale_factor""scale_factor""scale_factor""scale_factor", 'subsampling_step'"subsampling_step""subsampling_step""subsampling_step""subsampling_step", 'image_area'"image_area""image_area""image_area""image_area", 0.25, 0.5, 0.75, 1.0, 2, 3, 4

ObjectSampleIdxObjectSampleIdxObjectSampleIdxobjectSampleIdxobject_sample_idx (output_control) integer → (integer)

Index of the object sample.

Result

If the parameters are valid, the operator add_sample_identifier_training_dataadd_sample_identifier_training_dataAddSampleIdentifierTrainingDataAddSampleIdentifierTrainingDataadd_sample_identifier_training_data returns the value 2 ( H_MSG_TRUE) . If necessary an exception is raised.