Uploaded July 2026 | Updated September 2026, 10 minutes ago
This bachelor thesis presents terraClassiFly, an AI-based terrain classification pipeline developed in collaboration with Loft Dynamics. The system combines aerial imagery, near-infrared data, and elevation information to classify terrain into eight semantic categories using deep learning.
By relying only on globally available geospatial data, the approach enables realistic terrain interaction for helicopter flight simulators in regions where detailed land-cover maps are unavailable. The resulting terrain maps can be integrated directly into simulation environments, improving realism during takeoff, landing, and low-altitude flight.
This bachelor thesis presents terraClassiFly, an AI-based terrain classification pipeline developed in collaboration with Loft Dynamics. The system combines aerial imagery, near-infrared data, and elevation information to classify terrain into eight semantic categories using deep learning.
By relying only on globally available geospatial data, the approach enables realistic terrain interaction for helicopter flight simulators in regions where detailed land-cover maps are unavailable. The resulting terrain maps can be integrated directly into simulation environments, improving realism during takeoff, landing, and low-altitude flight.
