add_sample_class_knnT_add_sample_class_knnAddSampleClassKnnAddSampleClassKnn (Operator)

Name

add_sample_class_knnT_add_sample_class_knnAddSampleClassKnnAddSampleClassKnn — Add a sample to a k-nearest neighbors (k-NN) classifier.

Signature

add_sample_class_knn( : : KNNHandle, Features, ClassID : )

add_sample_class_knnadd_sample_class_knnAddSampleClassKnnAddSampleClassKnnAddSampleClassKnn adds a feature vector to a k-nearest neighbors (k-NN) data structure. The length of a feature vector was specified in create_class_knncreate_class_knnCreateClassKnnCreateClassKnnCreateClassKnn by NumDim. A handle to a k-NN data structure has to be specified in KNNHandleKNNHandleKNNHandleKNNHandleKNNHandle.

The feature vectors are collected in FeaturesFeaturesFeaturesFeaturesfeatures. The length of the input vector must be a multiple of NumDim. Each feature vector needs a class which can be given by ClassIDClassIDClassIDClassIDclassID, if only one was specified, the class is used for all vectors. The class is a natural number greater or equal to 0. If only one class is used, the class has to be 0. In case the operator classify_image_class_knnclassify_image_class_knnClassifyImageClassKnnClassifyImageClassKnnClassifyImageClassKnn will be used, all numbers starting from 0 to the number of classes-1 should be used, since otherwise an empty region will be generated for each unused number.

It is allowed to add samples to an already trained k-NN classificator. The new data is only integrated after another call to train_class_knntrain_class_knnTrainClassKnnTrainClassKnnTrainClassKnn.

If the k-NN classifier has been trained with automatic feature normalization enabled, the supplied features FeaturesFeaturesFeaturesFeaturesfeatures are interpreted as unnormalized and are normalized as it was defined by the last call to train_class_knntrain_class_knnTrainClassKnnTrainClassKnnTrainClassKnn. Please see train_class_knntrain_class_knnTrainClassKnnTrainClassKnnTrainClassKnn for more information on normalization.

Execution Information

Multithreading type: reentrant (runs in parallel with non-exclusive operators).
Multithreading scope: global (may be called from any thread).
Processed without parallelization.

This operator modifies the state of the following input parameter:

KNNHandleKNNHandleKNNHandleKNNHandleKNNHandle

The value of this parameter may not be shared across multiple threads without external synchronization.

Parameters

KNNHandleKNNHandleKNNHandleKNNHandleKNNHandle (input_control, state is modified) class_knn → (handle)

Handle of the k-NN classifier.

FeaturesFeaturesFeaturesFeaturesfeatures (input_control) number(-array) → (real)

List of features to add.

ClassIDClassIDClassIDClassIDclassID (input_control) integer(-array) → (integer)

Class IDs of the features.

Result

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

Possible Predecessors

train_class_knntrain_class_knnTrainClassKnnTrainClassKnnTrainClassKnn, read_class_knnread_class_knnReadClassKnnReadClassKnnReadClassKnn

References

Marius Muja, David G. Lowe: “Fast Approximate Nearest Neighbors with Automatic Algorithm Configuration”; International Conference on Computer Vision Theory and Applications (VISAPP 09); 2009.

Module

Foundation

Operators