search

intel / unet

U-Net Biomedical Image Segmentation
Apache License 2.0
306 stars 125 forks source link

Dice Score #32

Closed hc-codes closed 3 years ago

hc-codes commented 3 years ago

I successfully trained the 3d unet model for 30 epochs with a crop size of (64,64,64) and batch size 8 but the dice score was only 0.56. And testing avg dice was 0.4 Could you please help me solving this problem. And also please explain the usage of openvino. gita

tonyreina commented 3 years ago

You should try training for more epochs and possibly using a larger crop (I used 128x128x128) crops. That will yield better Dice.

OpenVINO is a framework for performing inference on Intel hardware that can significantly speedup your inference.