search

TimoSaemann / caffe-segnet-cudnn5

This repository was a fork of BVLC/caffe and includes the upsample, bn, dense_image_data and softmax_with_loss (with class weighting) layers of caffe-segnet (https://github.com/alexgkendall/caffe-segnet) to run SegNet with cuDNN version 5.
Other
176 stars 127 forks source link

Using Validation Dataset as a part of training Segnet #15

Closed Viswa14 closed 7 years ago

Viswa14 commented 7 years ago

Hi,

I wish to use val.txt file and validate my data during training process. If I have to add validation during training process, how do I handle Batch Normalization ? with respect to that what kind of changes are needed in the train.prototxt ?

TimoSaemann commented 7 years ago

Hi,

you don't need any changes for the batch normalization layers. You just need to change the first two lines of your solver file:

train_net: "/home/user/segnet_train.prototxt" 
test_net: "/home/user/segnet_val.prototxt"