search

BVLC / caffe

Caffe: a fast open framework for deep learning.
http://caffe.berkeleyvision.org/
Other
33.96k stars 18.72k forks source link

Accuracy for contrastive loss layer. #3809

Open cheer37 opened 8 years ago

cheer37 commented 8 years ago

I am training the siamese network using contrastive loss layer. But i see only the loss, so i dont know my net going well or not, and also how much. I want the feature of computing the accuracy for contrastive loss layer in caffe. Thanks.

seanbell commented 8 years ago

I agree that it would be helpful to add this, where accuracy is defined with respect to the contrastive loss threshold. Feel free to submit a PR implementing this feature.