UNET-MultiClass-Segmentation

segmentation of multiple classes

To run UNet- VGG

python3 train.py --epochs 50 --batch-size 4 --imsize 224 224 --classes 3 --augmentations --unet-vgg

• Import image size should be 224
• No need to use --amp , --dropout, --bilinear flag

To run UNet

python3 train.py --epochs 50 --batch-size 4 --imsize 968 732 --amp --classes 3 --dropout --augmentations --learning-rate 0.00001

• Use learning rate should be 0.00001