Terrain Classification Algorithm for Lunar Rover Using a Deep Ensemble Network with High-Resolution Features and Interdependencies between Channels

Table 9

Accuracy of the ensemble model.

Model	Train set	Test set	Val set
Model-1	92.61	91.07	89.75
Model-2	93.49	92.77	91.49
Model-3	94.11	93.16	91.54
Model-ensemble	94.31	93.24	91.57