bortizj / unet_keras_segmentation

Using Unet for image segmentation

Simple Unet example using keras for image segmentation

needs https://pytorch.org/get-started/locally/ pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

Epochs processed 50 / 50 ------------------Accuracy------------------ Class 0 Got 1602823/1672800 with accuracy 95.82 Class 1 Got 1648523/1672800 with accuracy 98.55 Class 2 Got 1668667/1672800 with accuracy 99.75 Class 3 Got 1615118/1672800 with accuracy 96.55 --------------------Dice-------------------- Class 0 Dice score 90.96 Class 1 Dice score 93.17 Class 2 Dice score 98.63 Class 3 Dice score 97.09

Epochs processed 500 / 500 ------------------Accuracy------------------ Class 0 Got 1613421/1672800 with accuracy 96.45 Class 1 Got 1653440/1672800 with accuracy 98.84 Class 2 Got 1669238/1672800 with accuracy 99.79 Class 3 Got 1622857/1672800 with accuracy 97.01 --------------------Dice-------------------- Class 0 Dice score 92.08 Class 1 Dice score 94.45 Class 2 Dice score 98.83 Class 3 Dice score 97.44

Performance training and validation set ------------------Accuracy------------------ Class 0 Got 5502636/5573280 with accuracy 98.73 Class 1 Got 5550683/5573280 with accuracy 99.59 Class 2 Got 5569571/5573280 with accuracy 99.93 Class 3 Got 5513671/5573280 with accuracy 98.93 --------------------Dice-------------------- Class 0 Dice score 97.25 Class 1 Dice score 98.13 Class 2 Dice score 99.64 Class 3 Dice score 99.07 Performance test set ------------------Accuracy------------------ Class 0 Got 591103/628320 with accuracy 94.08 Class 1 Got 613588/628320 with accuracy 97.66 Class 2 Got 620343/628320 with accuracy 98.73 Class 3 Got 603711/628320 with accuracy 96.08 --------------------Dice-------------------- Class 0 Dice score 89.28 Class 1 Dice score 89.62 Class 2 Dice score 94.27 Class 3 Dice score 96.05