Author's code for the paper "Set2Model networks: learning discriminatively to learn generative models"
To use the models in Caffe, please see the layers package: - MeanHistLayer - AVG baseline - GaussHistLossLayer - Set2Model-Gauss, Set2Model-GMM layers
To reproduce the experiments, please:
- obtain the image archive from the authors
- use the prepare.py script to fill the LMDB database
- use the corresponding caffe-format deep network specifications for training and testing, where substitute caffe-root for CCC, database directory for XXX, some directory to store the snapshots for TTT