I am a beginner in caffe framework and I would like to implement the training of a CNN model with multi-label data. I know how to create a lmdb data in python like this example but I couldn't find any complete tutorial of how to define the labels when an image has 2 or more labels.
Do you know if there is a complete and clear example of how to create such data and how to define the accuracy and loss layers in the prototxt files.
