Project Overview

The Kaggle Statoil/C-CORE Iceberg Classifier Challenge aimed to create a solution that distinguishes between icebergs and ships in satellite radar images. This challenge is critical for maritime navigation and environmental monitoring, helping to identify potential obstacles and contributing to Arctic and Antarctic studies.

Key Achievements

  • Top 5% Ranking: Outperformed thousands of entries by developing an effective solution that balanced accuracy and efficiency.
  • Optimized for Deployment: Achieved a highly compact model under 0.5MB, ideal for deployment in resource-limited environments.
  • Satellite Imagery Analysis: Tackled the challenges of interpreting radar images, where distinguishing features are often subtle.

Visual Insights

The satellite images revealed unique patterns that differentiate icebergs from ships. By focusing on these patterns, our classifier was able to interpret radar signal reflections and effectively categorize the objects.

Iceberg Signal

Iceberg Signal

Radar imagery showing the unique signal of an iceberg, essential for accurate classification.

Ship Signal

Ship Signal

Radar signal of a ship, highlighting differentiating features from icebergs.

Additional Resources

For more details on the Kaggle challenge, visit the Kaggle competition page

Updated: