Comparing classification accuracies using different neural network architectures (ResNet-50, VGG-19, Inception-v3, and a custom one) and two remote sensing datasets (SAT-6 and UCMerced).
Comparing:
1 - No augmentation
2 - Generic augmentation (cropping, rotatinon, flipping, etc.)
3 - Cross-correlation-based simulation algoright
SAT-6 data is located within its respective folder,
Due to data size, UCMerced dataset is stored in OneDrive:
https://uwy-my.sharepoint.com/:u:/g/personal/tdavydze_uwyo_edu/ETMAZAV_filAja1YfUSnBCwBYKB7huPNNDmDGW9UBHtUbg?e=ltKtP2