About the Project

Insects are vital members of the ecosystem, acting as vital pollinators for flowering plants, consumers of detritis, and in many cases keeping pests under control. However, some species – particularly ones non-native to a particular region – may have series detrimental effects by predating on native species or seriously damaging or destroying local plant life. Other insects such as mosquitoes and tsetse flies spread serious diseases such as malaria, dengue and yellow fever.

Whilst members of the public are encouraged to report sightings of harmful non-native species, correct identification is non-trivial for non-experts, resulting in many “false positive” sightings which can lead to the wasting of expert time and valuable resources. There is therefore need for automated tools for correctly identifying such harmful species, and the obvious ways of doing this is from images (e.g. which a member of the public might take on a mobile phone) or from the sounds they produce.

This project will involve application of Machine Learning approaches to classify digital images or sound recordings. The supervisors collaborate with expert entomologists, who can provide appropriate data and “ground truth” classifications.

The successful candidate will have good programming skills in at least one high level computer programming language, such as C/C++, Java or Python and have good understanding of Mathematical and Physical principles. An interest in Biology or Ecology would be an advantage.