This model is trained on the EuroSAT dataset to classify the 64x64 images into 10 classes namely -
- AnnualCrop
- Forest
- HerbaceousVegetation
- Highway
- Industrial
- Pasture
- PermanentCrop
- Residential
- River
- SeaLake
The CNN used is similar to AlexNet, with a dropout of 0.65 and 96 kernels instead of 64 in the first convolution layer.
The 27000 data samples were split into 20000 training samples and 7000 validation samples using a random split. The saved model classified the validation data with an accuracy of 96.4945%.
This model has been implemented using PyTorch Lightning.