BirdCLEF 2021 - Main task

Motivation

There are already many projects underway to extensively monitor birds by continuously recording natural soundscapes over long periods. However, as many living and nonliving things make noise, the analysis of these datasets is often done manually by domain experts. These analyses are painstakingly slow, and results are often incomplete. Data science may be able to assist, so researchers have turned to large crowdsourced databases of focal recordings of birds to train AI models. Unfortunately, there is a domain mismatch between the training data (short recording of individual birds) and the soundscape recordings (long recordings with often multiple species calling at the same time) used in monitoring applications. This is one of the reasons why the performance of the currently used AI models has been subpar.

To unlock the full potential of these extensive and information-rich sound archives, researchers need good machine listeners to reliably extract as much information as possible to aid data-driven conservation.

Data collection

The training set used for the challenge will contain approximately 60K recordings covering 397 species from North, Central and South America. Training recordings were collected by citizen scientists for the Xeno-canto network. The test set will contain soundscape recordings from four different recording locations: Eastern and Western United States, Costa Rica, and Colombia. Each recording is accompanied by metadata revealing the recording location which will help participants to evaluate location-dependent models. The validation data includes example soundscape recordings for two of the four recording sites.

Task description

Participants will have to train a classifier using short recordings of individual birds. For testing, the classifier will be applied to continuous soundscape recordings of varying complexity. Participants are asked to bridge the gap in acoustic domains between training and test recordings by crating models that generalize well across acoustic domains. Finally, a list of audible species for each five-second segment of soundscape data has to be retrieved.

How to participate ?

BirdCLEF 2021 will be held on Kaggle.

Click here to participate: BirdCLEF 2021 Competition Website

Prizes

The best scoring submissions will be awarded a cash prize:

1st Place - $2,500
2nd Place - $1,500
3rd Place - $1,000

We will also award the best working note:

Participants of this competition are encouraged to submit working notes to the BirdCLEF 2021 conference. As part of the conference, a best BirdCLEF working note competition will be held. The winner of the best working note award will be granted GCP cloud credit funds of $5,000.