USGIF’s GEOINT 2024 Symposium attracted a record number of GEOINT professionals to Kissimmee, Florida to explore the advancements in geospatial intelligence for the defense and space domains. This year’s theme was “Essential in All Dimensions and Domains”, which inspired conference sessions, panels, and presentations from the CIA, NGA, DoD, and other agencies highlighting both the potential and limitations of leading-edge technologies.
If you missed this year’s symposium, read on for all the biggest insights revealed onsite by the geospatial community. Or request your own demo of the Rendered.ai PaaS to see synthetic data generation for geospatial intelligence now!
Demonstrating the Power of the Rendered.ai PaaS at GEOINT 2024
The Rendered.ai team hosted an exhibit booth in the Carahsoft Pavilion, providing live demos and sharing how synthetic data generation powers innovation in AI, ML, and computer vision-based geospatial technology.
There was an overwhelming amount of expertise shared with our team from leaders in the private, public, and academic sectors over the four days of the conference. Many attendees visited the Rendered.ai booth eager to learn more about the advantages synthetic data generation provides for training artificial intelligence and machine learning systems, primarily in detection models for satellite imagery.
Mark Hogsett, Head of Federal Growth for Rendered.ai commented “ It was great to be at GEOINT 2024 and demonstrate how Rendered.ai’s physics-based synthetic data generation platform can seamlessly integrate into existing AI/ML workflows and deliver computer vision datasets at the pace and scale needed to train AI models used in dynamic environments.”
Interested in exploring the Rendered.ai PaaS for synthetic data generation and collaboration for computer vision projects? Activate a free trial!
The Rendered.ai Team’s Takeaways from GEOINT 2024
In our conversations with various visitors to the Rendered.ai booth and the many sessions we watched at GEOINT 2024, some key themes became apparent. Here are the Rendered.ai team’s takeaways from the conference:
1. Greater collaboration is needed between the government, commercial, and academic sectors to maintain and grow a technological edge.
This sentiment reverberated throughout all four days of the conference. On Day 2, Scott Bray, Assistant Secretary General for Intelligence and Security at NATO stated in his keynote, “The role and value of GEOINT is absolutely critical for our national and international security, and we need to stand together and work together as allies and like-minded partners to further build our GEOINT capabilities.” The Honorable John Sherman, CIO of the DoD emphasized that collaboration is the key to effectively applying cutting-edge technologies and adapting to emerging challenges in his keynote on Day 3. Sherman referenced the advancement of commercial imagery as a strong force for progress in geospatial intelligence and a further need for software modernization to enhance operational efficiencies.
Visitors to the Rendered.ai booth were shown how adopting a synthetic data generation Platform as a Service can help them to drive innovation, generate greater efficiencies, and improve collaboration to gain a competitive edge in training computer vision algorithms.
2. The importance of integrating computer vision and other AI capabilities into agency analytic work.
On Day 1, U.S. Navy VADM Frank Whitworth, Director of the National Geospatial-Intelligence Agency (NGA) delivered a captivating keynote providing updates on the Maven project, which integrates computer vision and other AI capabilities into the NGA’s analytic work. He highlighted five essential elements for applying AI at NGA:
-
- High-quality computer vision
- Integrating AI into the analytic workforce
- Assimilating AI into informed collection
- Implementing an efficient enterprise AI structure
- Leading the charge in the use of ethical, responsible AI principles
Scott Bray, Assistant Secretary General for Intelligence and Security at NATO said in his keynote on Day 2, “Spatial machine-learning algorithms and deep learning techniques have become extremely powerful, fueled by unprecedented computing power.” He went on to predict that using these technologies for large collections of imagery or geospatial data will help allies to improve intelligence production through applications like automated change detection in a variety of areas of interest, socio-economic analysis, maritime safety, analysis of space and cyber events, and more.
This theme was echoed in a panel on Day 3 by Brig. Gen. William Glaser, Director of the Synthetic Training Environment Cross Functional Team at Army Futures Command, Bob Pette, VP and General Manager of Enterprise Platforms at NVIDIA, and Michael Tschanz, Director, Engineering Technology and Analysis at The Walt Disney Company. Their panel explored many topics, such as the necessity of foundational geospatial models and digital twins to advance operations modeling and testing. Glaser noted that the U.S. must establish itself as the leader in AI and machine learning, although the nation is constrained by ethical and moral considerations that adversaries are not concerned by. Tschanz provided a valuable perspective from the commercial sector on the need for detailed simulations and digital twins to gaining a competitive edge. He cited these technologies as the key to how Walt Disney World is “going to transform how we develop attractions, immersive experiences, and keep pushing the boundaries.”
GEOINT 2024 participants saw how to leverage trusted synthetic datasets alongside simulation tools like DIRSIG, Blender, NVIDIA Omniverse and more in the Rendered.ai PaaS to provide easy integration of reiterative data generation into existing AI training workflows for edge cases and unusual scenarios. Leveraging a PaaS solution like this is an efficient solution to effectively integrate computer vision and AI capabilities into agency’s analytics programs.
3. Automation is beneficial but is no replacement for human-driven analytics.
The final main stage session on Day 2 of the GEOINT Symposium provided a forum for a panel of experts, including USGIF Board Chair Robert Cardillo, to discuss the opportunities and challenges associated with the geospatial intelligence tradecraft. Panel members stressed that automation has its place in geospatial intelligence but is not a replacement for analytics driven by humans. They exchanged ideas on keeping workforce development and training programs motivating and challenging, without a reliance on recruiting data-literate personnel.
This presents an interesting opportunity for organizations! Using a PaaS like Rendered.ai, gives agencies an inexpensive way to generate unlimited datasets to train personnel for standard situations, as well as more challenging edge cases and unusual field scenarios.
4. Navigating the diversity of data types and non-traditional data sources is difficult.
In her keynote on Day 1, the Honorable Stacey Dixon, Ph.D., Principal Deputy Director of National Intelligence pointed out that GEOINT is continually being redefined as technology and tactics progress, citing advancements in SAR, near-time imagery, and satellite constellations that track carbon output. Dixon also called out that with advances in AI and other emerging technologies, “Separating what is fake from what is real is becoming increasingly difficult.” It was also made apparent in the GEOINT 2024 Government Hub panel on May 6th featuring representatives from the CIA, U.S. Central Command, and the NGA that managing and merging disparate datasets in varying formats has become a great challenge for many agencies.
These presentations represent a call to arms for the technology industry to help build frameworks that the intelligence community can trust. For training and testing AI at the heart of geospatial technologies, this framework is found in a collaborative platform, like Rendered.ai, that gives organizations the ability to:
-
- Generate unlimited datasets rapidly for specific requirements, unusual scenarios, and edge cases for which real data does not yet exist
- Fuse and compare synthetic with real datasets leveraging 100% accurate data labeling
- Simulate SAR, RGB, multispectral, and hyperspectral imagery through trusted simulation tools like DIRSIG, Blender, NVIDIA Omniverse, and more
A synthetic data PaaS puts the power of versatile data generation, open team collaboration, and reliable imagery simulation in the hands of those who need it for mission-critical decision-making.
What’s Next?
These were all interesting insights from this year’s GEOINT Symposium with invaluable takeaways for the defense industry and beyond! The next step for exploring the benefits of synthetic data generation in the Rendered.ai PaaS to answer common data problems in AI and ML training is to try the Rendered.ai PaaS for yourself. Request a free trial of the platform today!
Learn more about the keynotes, panels, and other presentations at this year’s symposium at GEOINT 2024 News.
