SBIR-STTR Award

Open Call for Innovative Defense-Related Dual-Purpose Technologies/Solutions with a Clear Air Force Stakeholder Need
Award last edited on: 1/5/2022

Sponsored Program
SBIR
Awarding Agency
DOD : AF
Total Award Amount
$649,996
Award Phase
2
Solicitation Topic Code
AF192-001
Principal Investigator
Seth Potter

Company Information

AI.Reverie Inc

75 Broad Street Suite 640
New York, NY 10004
   (917) 697-7005
   N/A
   www.aireverie.com
Location: Single
Congr. District: 10
County: New York

Phase I

Contract Number: FA8649-19-P-A062
Start Date: 8/9/2019    Completed: 8/9/2020
Phase I year
2019
Phase I Amount
$50,000
AI.Reverie is a synthetic data simulation platform which generates an entirely new class of data that successfully trains and improves computer vision algorithms at scale. In order to train mission critical computer vision applications, there is a significant need for large amounts of diverse data that mimics real world locations and objects. Our company specializes in computer vision challenges - object detection and activity recognition - where data is highly sparse or non existent. AI.Reverie's sophisticated computer vision platform integrates machine learning and AI by generating synthetic data to help offset the costs of labeled real world data and images that are highly rare and unforeseen to mitigate risks associated with real world data biases. The company's technology trains the latest state of the art computer vision algorithms, which empowers drone warfare and keeps soldiers safe. AI.Reverie's platform automatically generates fully annotated synthetic data at scale (i.e., in the millions). The ability to synthetically generate data around scenarios that are highly dangerous, but are often difficult to find in the real world helps train algorithms that are better able to empower analysts in making life saving decisions.

Phase II

Contract Number: FA8649-20-9-9082
Start Date: 6/19/2020    Completed: 6/19/2021
Phase II year
2020
Phase II Amount
$599,996
AI.Reverie's simulation platform leverages the power of synthetic data to significantly improve the performance of mission critical vision algorithms. By generating fully annotated synthetic data at scale, the platform offsets the costs and time to label real world images and videos. The program objectives are focused on improving the accuracy of computer vision algorithms for object detection/tracking and activity recognition using fully annotated synthetic data for any type of sensor and at different altitudes (ground level to 64 square kilometers).