Seabird surveillance techniques

Ecology researchers may soon be able to collect large quantities of behavioural data on nesting seabirds using automated surveillance computer vision techniques.


Ecologists may soon be able to collect large quantities of behavioural data on nesting seabirds using automated surveillance computer vision techniques.



This is the hope of researchers at Microsoft Research’s Computational Ecology and Environmental Science group (CEES) who are working with ornithologists from the Evolutionary Biology and Behavioural Ecology group at SheffieldUniversity to develop automatic computer surveillance techniques for the monitoring of seabird populations.



The researchers at CEES have now teamed up with LincolnUniversity to further advance their computer vision algorithms as part of a project that has received funding from the EPSRC.



The teams aim to use these algorithms to monitor a population of Common Guillemots and automatically analyse video data of the birds in their cliff nesting area on SkomerIsland, off the coast of Wales. Researchers believe that this will provide a good indication of both seabird population trends and the health of marine life in general.



Existing methods use manual inspection to estimate the size of the population. However, this is a labour-intensive process and does not provide detailed data about individual birds, placing severe limitations on further analysis of their population.



According to the researchers, the investigation will allow ecologists to analyse new data on the relationship between chick survival and nest attendance. The collaboration with LincolnUniversity will focus mainly on the specific technical problems of automated visual surveillance and image processing in natural environments.



The researchers claim that while the project is focused primarily on the Skomer Guillemots, the proposed techniques could be deployed for monitoring other seabird species, and also used to support other applications of computer vision.



The £96,000 project is due to begin next month and will be completed by the end of November 2010.