search

sukritshankar / Caffe-LMDBCreation-MultiLabel

Creation of LMDB for training a multi-label loss in Caffe
71 stars 35 forks source link

Mirror should not exist in Train Data layer. Or should also exist in Train Label layer. #10

Open aniket03 opened 6 years ago

aniket03 commented 6 years ago

https://github.com/sukritshankar/Caffe-LMDBCreation-MultiLabel/blob/f4a92fcf4282ea808cd3c6bb5ade2712e23d2dbf/train_vgg_11_sigmoid_cross_entropy_loss.prototxt#L17

sukritshankar commented 6 years ago

The mirrored images should be carried forward with the same label as the original.