Deep Learning in Depth: IARPA's Functional Map of the World Challenge

Abstract

This was an image-recognition challenge on steroids. The problem was the United States has a lot of satellites that are fixed above the earth and has an overwhelming amount of satellite imagery of portions of the earths surface. What they are interested in doing is finding out what is going on on these plots of land. There are various functions that can be ascribed to buildings or facilities on the surface of the earth, like airport, amusement park, nuclear facility, or hospital. It is really important to be able to tell what is in this chunk of land for lots of intelligence-related reasons. There were several complications to the challenge that made it more interesting than just the standard, Does this photo have a dog or a cat in it? But, it was in spirit a very similar approach.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 2021
Accession Number
AD1128001

Entities

People

  • Carson Sestili
  • Ritwik Gupta

Organizations

  • Carnegie Mellon University

Tags

Communities of Interest

  • Space

DTIC Thesaurus Topics

  • Artificial Intelligence
  • Artificial Satellites
  • Data Science
  • Data Sets
  • Deep Learning
  • Department Of Defense
  • Detection
  • Engineering
  • Image Classification
  • Image Recognition
  • Information Science
  • Infrastructure
  • Learning
  • Machine Learning
  • Recognition
  • Satellite Imaging
  • Software Development
  • Standards
  • Terabytes
  • United States
  • United States Government

Readers

  • Educational Psychology
  • Neural Network Machine Learning.
  • Space Exploration and Orbital Mechanics.

Technology Areas

  • AI & ML
  • AI & ML - Machine Learning Algorithms
  • Space
  • Space - Orbital Debris