EuroSAT Satelitte Images Classification

Description

EuroSAT Satelitte Images Classification contains image classification models created to recognize type of land cover and land use. All models were trained with usage of fine tuning. Dataset comes from EuroSAT repository linked below.

Dataset source

Final models accuracies

DenseNet121

• Train accuracy: 0.9530
• Test accuracy: 0.9580

ResNet50v2

• Train accuracy: 0.9403
• Test accuracy: 0.9592

Xception

• Train accuracy: 0.9486
• Test accuracy: 0.9565

VGG19

• Train accuracy: 0.7793
• Test accuracy: 0.8781

Technologies used

• TensorFlow
• Keras
• Python