Australian Geospatial-Intelligence Organisation (AGO) Analytics Labs

Addressing AGO capability challenges through a small number of short-term industry projects, with a focus on machine learning and analytics for producing automated imagery analysis.

The Challenge

The Australian Geospatial-Intelligence Organisation (AGO) often deals with sensitive and classified problems, and as a result, the range of companies and professionals they can access is much smaller than the broader pool of skills and capabilities available in the market. This often makes it challenging for AGO to work with new organisations, and particularly with small to medium enterprises who have had no previous track record of working with Defence. The AGO is keen to attract a wider pool of companies and technologies to draw on for automated geospatial intelligence. The primary focus of the program is to address AGO capability challenges through a small number of short-term industry projects, with a focus on machine learning and analytics for producing automated imagery analysis, including automated object classification.


In 2019, the AGO approached FrontierSI with this challenge. FrontierSI proposed a collaborative, practical Analytics Labs Innovation Program (AGO Labs) to solve these challenges. The approach was for FrontierSI to work with AGO stakeholders to identify and articulate three geospatial challenges common to Defence, which could be generalised and made available in an unclassified way. These challenges were then released in an open call to Australian private industry and universities, and companies were invited to submit innovative approaches to solve these challenges.

The Solution

Four of the most novel approaches to the three challenges were selected, granted up to $100,000 in funding, and FrontierSI and AGO collaborated with the successful organisations to deliver innovative challenge responses as a series of parallel, short 6-month, projects. A brief description of each of the four projects and their enabling activities follows:

  • Featured Unsupervised (organisation: Ozius). Ozius’ Deep Insight AI approach demonstrated advanced decision-making capabilities by the rapid detection of change types and labelling of anomalous anthropogenic and environmental features across large areas.
  • Low-cost Object Identification Models (organisations: Urbis and the Australian Institute of Machine Learning, University of Adelaide). This project demonstrated efficiencies in object detection in the machine learning training process, particularly focused on rare features. The project demonstrated a new method to self-generate training features, leading to more accurate models using less training data.
  • Low-cost Object Identification Models (organisation: Microsoft Australia). The above challenge was approached in a separate manner. Microsoft deployed a novel perspective in generating synthetic data sets which both dramatically reduce the requirement for real training data and improve the efficacy of machine learning models faster, cheaper, and more accurately than relying on generic data sets.
  • Beautiful Contours (organisations: Geospatial Intelligence and Maxar Technologies). This project aimed to automate the complex task of creating cartographic quality contours from diverse elevation datasets in near real time, suitable to meet the needs of operational users anywhere on the globe.

FrontierSI is continuing AGO Analytics Labs in 2023, with a Data Generalisation capability demonstrator challenge.


AGO received exposure and easy access to a new pool of Australian industry and academic organisations and their associated capabilities from which to draw upon for technology support in automated geospatial intelligence. Through the successful conduct of four technology showcase projects, AGO Labs has demonstrated that Australian industry possesses both technologies and capabilities that can assist AGO in solving existing defence challenges as well as identifying new challenges.

Key learnings for AGO included a greater understanding of the strengths of local industry AI/ML capability, a proven mechanism for industry engagement and internal problem definition, improved AGO staff awareness of current leading AI/ML approaches and technologies, and an improved AGO ability to work with small business who are unfamiliar with the objectives and processes of Defence and intelligence organisations. Additional outcomes for AGO were an appreciation of the enthusiasm that companies new to the defence intelligence arena could bring to the environment and the benefits that an innovative approach could have in unexpected but valuable ways.

AGO identified that the partnership with FrontierSI enabled them to do more, faster and better, and that the value delivered was much greater to the cost of the program. The program was regarded as highly successful.


To learn more, contact FrontierSI at, or connect with Project Manager Roshni Sharma at