Skip to main content

Computational Methods for Decision Making – Automated Image Understanding


The Computational Methods for Decision Making applied research program is composed of three thrusts — Large Scale Distributed Decision-making, Automated Image Understanding, and Resource Optimization.

The Automated Image Understanding thrust focuses on further developing promising computational methods based on principled approaches for building reliable vision systems. The primary focus is on high-level vision:

  • Recognizing objects and activities, and inferring intentions and threats in surveillance imagery
  • Perception for autonomous agents
  • Semantic search of large image and video databases

This program is applied research and focuses on naval problems.


Research Concentration Areas

  • Methods for integration of high-level knowledge and image data, in particular hierarchical and compositional models for representation and processing, and reasoning
  • Learning complex visual concepts, in particular learning relationships between objects, activities involving interactions of people and objects/spaces, etc.
  • Using non-visual information to provide additional context for recognition and inference
  • Reconstruction of “scenes and events” from ad hoc networks of heterogeneous cameras, where cameras may be still or video, stationary or mobile, with different resolutions and image quality, and with images taken from widely different views
  • Methods for analyzing images from other modalities and fusion of multimodal imagery for enhanced recognition

Research Challenges and Opportunities

  • Task relevant learning and visual representations of scenes, objects, activities, events
  • Architectures and processes for integration of high-level visual knowledge and image/video data for recognizing scenes, objects, activities, and events
  • Dynamic scene understanding from multiple cameras, other sensing modalities, and ancillary information

How to Submit

For detailed application and submission information for this research topic, please refer to our broad agency announcement (BAA) No. N0001425SB001.

Contracts: All white papers and full proposals for contracts must be submitted through FedConnect; instructions are included in the BAA.

Grants: All white papers for grants must be submitted through FedConnect, and full proposals for grants must be submitted through grants.gov; instructions are included in the BAA.


PROGRAM CONTACT INFORMATION

Name
Dr. Behzad Kamgar-Parsi
Title
Program Officer
Department
Code 311