A PyTorch implementation of UNet++: A Nested U-Net Architecture for Medical Image Segmentation trained with CelebAMask
I made new datasets in Kaggle after pre-processing.
optimizer: Adam
learning rate: 3e-4 (without scheduler)
image size: 128x128
epoch: 100
batch size: 12
train loss: 0.5 * binary cross entrophy + dice coefficient loss
validation & test metric: IoU (Intersection over Union)
Some outputs are even better than target images in the dataset.
Train | Valid | Test | |
---|---|---|---|
# of images | 27000 | 2500 | 500 |
IoU | 0.9892 (Epoch 100) |
0.9616 (Epoch 93) |
0.9618 |