AvesEcho
AvesEcho is a species classifier for bird audio recordings, specifically developed to target the bird species audible in continental Europe. Users upload a short sound clip, and it returns its list of estimated species, each with a confidence measure that can be interpreted as a kind of probability. For the case studies, bulk analysis of passive acoustic monitoring (PAM) sound collections is implemented, to investigate the patterns of occurrence of bird vocalisations.
Basic Technologies Used
Deep Learning
Convolutional Neural Networks
Transformers
Transfer Learning
Usefulness in GUARDEN Case Studies
NCC field team is responsible for the avian post-construction monitoring project for the windfarms established in the area of Karystos, at the southern tip of the island of Euboea, close to Athens, for the next 3 years. We are investigating bird activity in relation to the placement of the wind turbines, by automatically detecting the presence of birds detected from a set of fixed sound recording points.
The Cyprus case study concerns the expansion of an energy centre. The sensors (in natural areas and near agricultural fields) are positioned to measure bird vocal activity in the area, and to investigate how these are affected by land-use and other local factors.
Contact Information
- Burooj Ghani, Naturalis
- burooj.ghani@naturalis.nl