about the train.prototxt and deploy.prototxt

[abcdvzz]

When I use your delpoy.prototxt, do i need to add a fc layer after the last conv layer? or i just need a softmax layer

[farmingyard]

@abcdvzz The deploy is for imagenet 1000 classification, so I have fc1000 layer, but if you do other classification task, you may change to your own fc layer. Besides, you also need an input data layers and softmax layer for training.