The Intelligence Advanced Research Projects Activity (IARPA) recently opened registration for a prize challenge in automated analysis and classification of objects within satellite imagery.
IARPA’s Functional Map of the World (fMoW) Challenge asks developers to build machine learning algorithms capable of classifying the function of facilities, specific buildings, and land—information used by intelligence personnel to support defense, humanitarian, and disaster response missions.
“We are hoping to introduce learning opportunities for the geospatial and deep learning communities to integrate their approaches and increase the exposure to the scientific gains that could be made by combining these two disciplines,” said IARPA spokesperson Charles Carithers.
Labeling objects within satellite imagery is time-consuming when handled by a human analyst and contributes to operator burnout. This challenge aims to produce breakthroughs in deep learning analysis that will accelerate this process and, in the absence of human error, improve accuracy.
In July and August, IARPA will release in two batches one of the world’s largest publicly available satellite image data sets to date, including roughly one million image “chips.” Each chip will highlight an unidentified point of interest and the library will contain 62 pre-defined categories. This data is the fuel participants will feed their algorithms.
“The release of this data set will stimulate further research and learning throughout the computer vision community to include hobbyists, academia, and industry,” Carithers said. “The availability of the data will enable development of solutions by those who don’t have easy access to this type of annotated data.”
Finalists will be selected based on correct classification of known points of interest from this data set. The most effective algorithms will be evaluated for speed and precision to determine final rankings.
Participants will compete for $100,000 in total prizes, including a $25,000 first place prize. Experts, academics, and novices alike are welcome to compete.
Those interested in participating may register now. Final submissions are due in December and awards will be announced in February 2018.
Can you build algorithms to classify facility, building, and land use from satellite imagery?