The Defence Science and Technology Laboratory’s (Dstl) data science competition on Kaggle attracted over 5,000 submitted solutions. This equates to over £2.5m of research - a seven fold return on the cost of running the competition.
Commenting on the competition Dstl’s Phil Gibson said:
“The competition has been very well received by the Kaggle community. It just goes to show that the ‘Kagglers’ deserve their reputation for coming-up with exciting, innovative and thought provoking solutions. The fantastic response shows that crowdsourcing solutions can be an invaluable tool in tackling complex challenges – helping us secure the nation’s defence, security and prosperity.”
We are now analysing the data. The algorithms and software will be used in a number of projects where the ability to rapidly and accurately analyse data sets is important.
In a UK Government first, Dstl worked with Kaggle, the world’s largest data science competition community, to identify novel methods to evaluate large, complex data sets. This will help intelligence analysts evaluate information more quickly, accurately and effectively.
Dstl challenged users to develop an algorithm or software that will automatically detect and identify objects in satellite images. The challenge was to correctly identify and categorise up to 10 features and objects, such as cars, trees and buildings in a variety of environments. The project has the potential to identify solutions that will greatly improve the UK’s satellite intelligence gathering and analysis capability.
Details of the winning entries can be found on Kaggle’s website.