skip connected retrieval network
download data and extract it to ./data/datasetname
change the hyper_perameter for the corresponding data,ie.dataset name, class number, and etc.
run train.py
When training is completing, run evaluate.py to test the network.
Further information is on the way