ARLINGTON, Va. – U.S. military researchers will brief industry on 11 March 2022 on the upcoming Fiddler program (DARPA-SN-22-23) to improve automatic object recognition in synthetic aperture radar (SAR) images. Briefings will be via Microsoft Teams, and will be from 1 to 4 p.m. eastern time.
Commercial and government investments are leading to rapid growth in Earth-observation satellites and remote-sensing data, DARPA researchers say. In particular SAR can produce high-resolution images of the Earth at night and in all-weather conditions.
This unique imaging capability makes SAR particularly useful for time-critical applications like change-detection after natural disasters and identifying illegal fishing operations.
The objective of the Fiddler program is to improve automatic object recognition in SAR images. Object recognition often requires significant examples to train machine learning classification algorithms. Obtaining training data can be time consuming, expensive, and even impossible in dynamic conditions.
The use of machine learning and computer vision methods to generate training data in dynamic maritime environments is of particular interest to this program.
Performers first will develop ways to create object reference models directly from real SAR image examples. From these models, they will develop how to generate or render synthetic SAR images of the object at new imaging geometries and configurations.
Performers then will demonstrate generation of diverse training data to train robust SAR object detection methods rapidly from few real examples.
Companies interested should register for the online briefings no later than 4 March 2022 online at https://cvent.me/01Bgy9.
Email questions or concerns to Kevin Rudd, the DARPA Fiddler program manager, at [email protected]. More information is online at https://sam.gov/opp/94f14230d6424489a5839383e6833d24/view.