The National Geospatial-Intelligence Agency has kicked off a prize competition that challenges participants to submit machine learning-based approaches to determine the location of audio or video recordings.
NGA said the five-week submission period for its Soundscapes competition runs through Nov. 27 and the agency will assess each proposed geolocation method based on its potential to identify, analyze and model scene indicators of ambient sound.
Participants will aim to develop algorithms that locate the city where NGA-provided audio files have been recorded.
The agency plans to award $27K to the contest's grand winner and a total of $10K to the top two entries from the academic community.
Top scorers will have the chance to present their methodologies during a workshop the NGA will host next year.