Enabled Intelligence Case Study: 95% to 99% Accuracy in Geospatial AI

Challenge

Enabled Intelligence, a full-stack AI technology and service provider specializing in geospatial analysis, faced the complex challenge of building reliable AI models for mission-critical applications. As one of the leading experts in geospatial AI in the US, the company needed to process thousands of geospatial images and create millions of accurate labels for diverse use cases spanning military intelligence, disaster recovery, transportation logistics, and agricultural monitoring.

Their projects required the highest quality data.

• Geospatial analysis for government and commercial clients demands extremely accurate data labeling
• Ensure at least 95% accuracy across all training data, with some ground truth datasets requiring 99% accuracy or higher.
• Technical complexity multiplied the challenge: The team had to handle coordinate system transitions from geospatial coordinates for labeling to pixel coordinates for AI training, then back to geospatial coordinates for deployment.
• Datasets varied: multiple sensor types including synthetic aperture radar, electro-optical imagery, hyperspectral data, and full motion video

The company needed a platform that could handle the unique requirements of geospatial data while supporting their rigorous quality assurance workflows essential for government and commercial security standards.

Solution

After evaluating over 35 different labeling platforms and software packages, Enabled Intelligence selected Kili Technology as their primary labeling platform, integrating it as a core component of their open architecture data annotation system.

Kili Technology's Impact:

• Kili Technology's platform proved uniquely suited for geospatial data labeling challenges.
• One of Kili's standout features for geospatial applications is its measurement capability. Enabled Intelligence uses this tool to measure objects in real-world coordinates
• Kili's image layering capability enables sophisticated analysis workflows. The platform allows stacking multiple images—such as synthetic aperture radar imagery with electro-optical imagery—and toggling between layers
• Enabled Intelligence implemented rigorous quality control processes using Kili's flexible workflow capabilities
• Built around Kili's platform, Enabled Intelligence tracks comprehensive accuracy metrics including false positive rates, false negative rates, misclassifications, and label alignment accuracy

Get access to the full webinar

Outcome

Enabled Intelligence's implementation of Kili Technology has established them as a premier geospatial AI data provider, delivering mission-critical accuracy for the most demanding applications.

Results

• After comprehensively evaluating 35+ labeling platforms, Kili Technology emerged as the clear leader for geospatial applications, becoming the foundation of Enabled Intelligence's data annotation capabilities across multiple sensor types and use cases.
• The platform enables Enabled Intelligence to consistently deliver 95% accuracy across all training data, with ground truth datasets achieving 99%+ accuracy for AI model testing and evaluation
• Enabled Intelligence now processes thousands of geospatial images and creates millions of labels across diverse applications including military aircraft detection, naval vessel identification, agricultural monitoring, disaster recovery analysis, and land use classification
• Enabled Intelligence continues expanding their geospatial AI capabilities, building both foundational models for broad applications and fine-tuned expert models for specific use cases, while maintaining their library of pre-labeled datasets for rapid deployment

"Kili is the labeling platform that we use and has been the best that we have found for labeling geospatial data."

Peter Kant
‍CEO at Enabled Intelligence