You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
""" Then drop non-dominant subcenters and high-confident noisy data, which is `>75 degrees` """importdata_drop_top_k# data_drop_top_k.data_drop_top_k('./checkpoints/TT_mobilenet_topk_bs400.h5', '/datasets/faces_casia_112x112_folders/', limit=20)new_data_path=data_drop_top_k.data_drop_top_k(tt.model, tt.data_path)
""" Train with the new dataset again, this time `loss_top_k = 1` """tt.reset_dataset(new_data_path)
optimizer=tfa.optimizers.SGDW(learning_rate=0.1, weight_decay=5e-4, momentum=0.9)
sch= [
{"loss": losses.ArcfaceLoss(scale=16), "epoch": 5, "optimizer": optimizer},
{"loss": losses.ArcfaceLoss(scale=32), "epoch": 5},
{"loss": losses.ArcfaceLoss(scale=64), "epoch": 40},
]
tt.train(sch, 0)
Train TopK 1 from BottleneckOnly
Train with initial_epoch = 0, that learning_rate will be [0.1, 0.01, 0.001]
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
-
Default import
Baseline
SubCenter ArcFace
Train TopK 1 from BottleneckOnly
initial_epoch = 0
, thatlearning_rate
will be[0.1, 0.01, 0.001]
initial_epoch = 40
, thatlearning_rate
will be0.001
Result and Plot
Beta Was this translation helpful? Give feedback.
All reactions