train.py --epochs 1 --weight_decay 1e-3 --lr 1e-5 --batch_size 64 --network 1 --view 6 --alpha 1.0
test.py --weight_decay 1e-3 --batch_size 64 --network 3 --view 6 --tm 336385
1:Frontal 2:Lateral 3:DualNet 4:Stacted 5:Our method without Auloss and mimicry loss 6:Our method with Auloss and mimicry loss 7:HeMIS
1:Resnet 2:Xception 3:Densenet