To tackle these crippling global diseases requires targeted control of their vector, the mosquito. This in turn relies upon detailed knowledge of the distribution, diversity and abundance of mosquitos in space and time.
Current mosquito survey methods are time consuming, expensive, spatially limited and can put those conducting them at risk of catching the vector borne diseases they are trying to prevent. Consequently, there is an urgent need to find new, automated and reliable survey methods.
The Humbug project required algorithms that could detect and identify different species of mosquitoes using the acoustic signature (sound) of their flight tones captured on a smartphone. Equally, this presented a challenge of scale; for the approach to be effective, the models had to be easily made accessible to many phones.
Mind Foundry helps the HumBug project by providing state-of-the-art machine learning models for the classification of mosquito sounds, and by developing new algorithms.
The Mind Foundry Platform provides a tool to experiment with different detection techniques, as well as a scalable platform for deploying predictive models that can be called in a distributed fashion by thousands of mobile phone at the same time.
In house expertise in signal processing and wavelet decomposition led to a strong predictive set of features, which were combined with Mind Foundry’s model building and optimisation technology. This quickly delivered a solution that outperformed standard approaches and modelling techniques, delivering more accurate and comprehensive detection of mosquitoes in audio recordings, even outperforming human experts.
Mosquito sounds recorded by a smartphone could be analyzed by an AI detection algorithm.
This crowdsourced method for tracking the spread of malaria is faster and more accurate than other methods currently used.
Performance is faster and more accurate at detecting the presence of mosquitoes.
Learn more at humbug.ox.ac.uk/