Train UNet with 100 images

shuaigezhu / star_galaxy_classification

0 stars 1 forks source link

Closed kmyi closed 4 years ago

kmyi commented 4 years ago

Use Fully convolutional version (the padding options should be set so that you use valid convolutions only within the network)

shuaigezhu commented 4 years ago

Hi, Kwang. I don't quite understand what the meaning of 'only within the network'.

kmyi commented 4 years ago

Sorry what I meant was "use valid convolutions only", within the network