milesial / Pytorch-UNet

PyTorch implementation of the U-Net for image semantic segmentation with high quality images
GNU General Public License v3.0
9.01k stars 2.47k forks source link

dice = ? 0.5 #480

Open zx0412 opened 7 months ago

zx0412 commented 7 months ago

why Validation Dice score always equals 0.5?

Validation round: 43%|████▎ | 13/30 [00:06<00:03, 5.47batch/s] Validation round: 50%|█████ | 15/30 [00:06<00:02, 6.65batch/s] Validation round: 70%|███████ | 21/30 [00:06<00:00, 14.40batch/s] INFO: Validation Dice score: 0.5 Epoch 14/50: 30%|███ | 84/277 [00:18<00:08, 23.33img/s, loss (batch)=0.00319] Validation round: 0%| | 0/30 [00:00<?, ?batch/s] Validation round: 3%|▎ | 1/30 [00:02<01:22, 2.86s/batch] Validation round: 7%|▋ | 2/30 [00:04<00:51, 1.85s/batch] Validation round: 10%|█ | 3/30 [00:04<00:30, 1.15s/batch] Validation round: 13%|█▎ | 4/30 [00:04<00:23, 1.08batch/s] Validation round: 17%|█▋ | 5/30 [00:05<00:17, 1.41batch/s] Validation round: 20%|██ | 6/30 [00:05<00:15, 1.57batch/s] Validation round: 23%|██▎ | 7/30 [00:05<00:11, 1.94batch/s] Validation round: 30%|███ | 9/30 [00:06<00:07, 2.66batch/s] Validation round: 33%|███▎ | 10/30 [00:06<00:06, 3.19batch/s] Validation round: 47%|████▋ | 14/30 [00:06<00:02, 6.66batch/s] Validation round: 57%|█████▋ | 17/30 [00:06<00:01, 9.05batch/s] Validation round: 67%|██████▋ | 20/30 [00:06<00:00, 11.57batch/s] INFO: Validation Dice score: 0.5 Epoch 14/50: 50%|█████ | 139/277 [00:30<00:05, 23.38img/s, loss (batch)=0.0219]