tracktable.domain.feature_vectors module

Module contents

tracktable.domain.feature_vectors.convert_to_feature_vector(coords)

Package up a list of numbers as a feature vector

Conceptually, a feature vector is just a list of numbers. However, in Tracktable, the functions that work with feature vectors (DBSCAN and the R-tree) expect specific data structures in order to go back and forth between Python and C++. Specifically, this data structure is a tracktable.domain.feature_vectors.POINT_TYPES[NN], where NN is the number of values. Right now NN can be no greater than 30.

The function convert_to_feature_vector() takes a list or tuple of numeric values and packages them up in that data structure. It does not change or reorder the values in any way.

Parameters:

coords (list or tuple of numbers) – Sequence of feature values

Returns:

Input values bundled up in a FeatureVectorNN data structure as

described above

Note

In the future, it would be great if the user never had to care about this function – feature vectors would get automatically instantiated when the R-tree is built or DBSCAN is invoked.