DCASE 2021 Task 5: Few-shot Bioacoustic Event Detection Evaluation Set

General Description

The evaluation set for task 5 of DCASE 2021 "Few-shot Bioacoustic Event Detection" consists of 31 audio files acquired from different bioacoustic sources. 

In Evaluation_Set_Annotations: the first 5 annotations are provided for each file, with events marked as positive (POS) for the class of interest. This is the same setup used during the DCASE 2021 challenge.

In Evaluation_Set_Full_Annotations: the full annotations are provided for each file, with events marked as positive (POS) or unknown (UNK) for the class of interest.

Folder Structure

Evaluation_Set.zip contains audio files and annotation files with 5 first POS events (as used during DCASE 2021 challenge)

|__Evaluation_Set/

    |___DC/

        |____*.wav

        |____*.csv

    |___ME/

        |____*.wav

        |____*.csv

    |___ML/

        |____*.wav

        |____*.csv

Evaluation_Set_Audio.zip has the same structure but contains only the *.wav files.

Evaluation_Set_Annotations.zip has the same structure but contains only the *.csv files with first 5 POS annotations.

Evaluation_Set_Full_Annotations.zip has the same structure but contains only the *.csv files with all POS annotations.

The subfolders denote different recording sources and there may or may not be overlap between classes of interest from different wav files.

Annotation structure

Each line of the annotation csv represents an event in the audio file. The column descriptions are as follows:
[ Audiofilename, Starttime, Endtime, Q ]

Classes

DCASE2021_task5_evaluation_set.csv provides a table with class code correspondance to class name for all the recordings of the Evaluation set.

DCASE2021_task5_evaluation_set.csv
-------------------
dataset, recording, class_code, class_name

Development Set

The development set for the same task can be found at: https://doi.org/10.5281/zenodo.5412896

Open Access

This dataset is available under a Creative Commons Attribution 4.0 International (CC BY 4.0) license.
 

Contact info

Please send any feedback or questions to:
Veronica Morfi: g.v.morfi@qmul.ac.uk