Call for participation: Cross-Domain Mosquito Species Classification Challenge

[H YB]

Call for participation:
BioDCASE 2026 Cross-Domain Mosquito Species Classification Challenge

Jointly organised by teams at the University of Oxford, King’s College London, and the University of Surrey, this challenge focuses on a key real-world question:

Can mosquito species classifiers still work when recordings come from new locations, devices, and acoustic environments?

Mosquito-borne diseases affect over 1 billion people each year. Audio-based monitoring could help scale surveillance, but domain shift remains a major barrier to real-world deployment.

To support transparent and reproducible research, we are releasing:

• an open development dataset with 271,380 clips and 60.66 hours of audio;
• a fully public, lightweight baseline that is easy to run;
• a benchmark focused on cross-domain generalisation in mosquito bioacoustics.

Participants are warmly invited to join and help develop more robust methods for mosquito monitoring under real recording conditions.

Useful Links:

• Challenge Website: [https://biodcase.github.io/challenge2026/task5]
  Baseline code: [https://github.com/Yuanbo2020/CD-MSC]
  Dataset: [https://zenodo.org/records/19095788]

Key Dates:
• April 1, 2026: Challenge opening
• Jun 1, 2026: Evaluation set release
• June 15, 2026: Challenge submission deadline

Feel free to share this with anyone who might be interested!

Apologies for cross-posting.
Best wishes,
 
Yuanbo