Hi,
Recently a tutorial was added on how to do Multilabel classification on PASCAL using python data-layers. (
http://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/pascal-multilabel-with-datalayer.ipynb)
Here, the data is input using python data layers using the wrappers for Pascal dataset. The tutorial says that this can be replaced by lmdb and hdf5 formats. Could you please share how this can be done?
Also, can a different dataset (say something like MNIST) be used to show how multilabel classification can be done using python data layers?
Thank you.