SBIR-STTR Award

TOFENet Topographic Features Extraction Network
Award last edited on: 3/8/2024

Sponsored Program
SBIR
Awarding Agency
DOD : NGA
Total Award Amount
$1,100,000
Award Phase
2
Solicitation Topic Code
NGA181-001
Principal Investigator
Kyle Ashley

Company Information

Intelligent Automation Inc (AKA: IAI)

15400 Calhoun Drive Suite 190
Rockville, MD 20855
   (301) 294-5200
   contact@i-a-i.com
   www.i-a-i.com
Location: Single
Congr. District: 06
County: Montgomery

Phase I

Contract Number: HM047618C0046
Start Date: 9/6/2018    Completed: 6/15/2019
Phase I year
2018
Phase I Amount
$100,000
Topographic features found in ground-based natural images contain information that is useful for a variety of applications including location estimation and navigation. Traditionally these features have been manually labeled by analysts which is costly and time consuming, especially considering the volume of readily available data. We propose a novel method for extracting topographic features from still images and video using a state-of-the-art deep learning based approach. Cues gathered from RGB images are used to locate topographic features in individual images and include information about the relative size, position, and context of each feature. Furthermore, by tracking these features within frames of a video the system can reliably mitigate false positives and gather a more comprehensive understanding of the environment being imaged. Using carefully chosen training datasets our approach learns to estimate features such that the output is robust to changes in lighting,weather and environment. The proposed system will provide a pixel-wise labelling of topographical features, along with useful contextual metadata about each feature.

Phase II

Contract Number: HM047619C0092
Start Date: 9/26/2019    Completed: 9/29/2020
Phase II year
2019
Phase II Amount
$1,000,000
Topographic features found in ground-based natural images contain information that is useful for a variety of applications including geolocation estimation and navigation. Traditionally, these features have been manually labeled by analysts which is costly and time consuming, especially considering the volume of readily available data. During Phase I, we have demonstrated a novel TOFENet framework for extracting topographic features from still images using a state-of-the-art deep learning based techniques. In this Phase II, we propose to extend the capability to video data, improve robustness to handle variations in lighting, weather and environment, and reduce false positives. IAI will build and deliver a TOFENet software ready for integration, by the end of Phase II.