System for classifying events in the vicinity of an optical fiber (event_classifi

Authors

Ing. Zoltán Galáž, Ph.D., Ing. Jiří Mekyska, Ph.D., doc. Ing. Vít Novotný, Ph.D., doc. Ing. Petr Sysel, Ph.D., Ing. Jiří Přinosil, Ph.D., Ing. Vojtěch Zvončák

Date

15. april 2021

Description

The event_classifier is a pip-installable Python package for classifying events that occurred in the vicinity of an optical fiber. It uses an ensemble of Gaussian mixture models that classify an unknown event into the mixture of interest utilizing features based on Mel-spectrograms. It provides users with an option to easily add/remove new events and setting the configuration file to support such a flexibility. The package also comes with a set of Jupiter notebooks for feature extraction and re-training of the models.

Event classifier (high level)
compression

Projects

This work was supported by project VI20172020078.

License

The software was developed within a research project at BUT and is further provided according to the license conditions on the basis of an agreement on the use of joint results. Contact: doc. Ing. Vít Novotný, Ph.D. (novotnyv@fbm.vutbr.cz).