is possible to run load_fast_files() in train_2ddense.py?

xmengli / H-DenseUNet

TMI 2018. H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation from CT Volumes

526 stars 161 forks source link

Closed dzfowen closed 5 years ago

dzfowen commented 5 years ago

It seems to load 131 volumes once in RAM, which needs optimization and cannot run in a normal server.

xmengli commented 5 years ago

Loading 131 volumes once in RAM can save training time. Alternatively, you can load image path only and read images when training the network.

dzfowen commented 5 years ago

thanks what about 131 lines liverlist[] meaning in train_2ddense.py?

xmengli commented 5 years ago

The index in liverlist is the data case that does not contain tumor.