This repository contains documentation and code for the GEO lecture .
-
Bu normal fontlu bir cümle.
-
Bu italic fontlu bir cümle.
-
Bu da italic fontlu bir cümle.
-
Bu kalın fontlu bir cümle
-
Bu da neden kalın fontlu bir cümle olmasın?
def forward(self, x):
"""Sequentially pass `x` trough model`s encoder, decoder and heads"""
features = self.encoder(x) # Extract features
decoder_output = self.decoder(*features) # Decoder output for segmentation task
pred = self.segmentation_head(decoder_output) # Feed the decoder output to the segmentation head
reconstruct_decoder_output = self.skipless_decoder(*features) # Decoder output for reconstruction task
reconst_mask = self.reconstruct_segmentation_head(reconstruct_decoder_output) # Feed the decoder output to the reconstruction head
return pred,reconst_mask
Dataset used in this work is from SpaceNet6 challange. See the arXiv paper here: "SpaceNet 6: Multi-Sensor All Weather Mapping Dataset"
precision recall f1-score support
building 0.90 0.86 0.88 38508672
not_building 0.99 0.99 0.99 547096448
accuracy 0.98 585605120
macro avg 0.95 0.93 0.94 585605120
weighted avg 0.98 0.98 0.98 585605120