search

TreB1eN / InsightFace_Pytorch

Pytorch0.4.1 codes for InsightFace
MIT License
1.72k stars 418 forks source link

Has anyone tried other backbone networks for training? #162

Open DJMeng opened 3 years ago

DJMeng commented 3 years ago

Such as：mobilenet-v2, I trained 40 epochs,The loss dropped very slowly, only from 30 to 25.

Tamminhdiep97 commented 3 years ago

Such as：mobilenet-v2, I trained 40 epochs,The loss dropped very slowly, only from 30 to 25.

I Tried on EfficientNet-B4, loss also dropped so so so slow.

DJMeng commented 3 years ago

Such as：mobilenet-v2, I trained 40 epochs,The loss dropped very slowly, only from 30 to 25.

I Tried on EfficientNet-B4, loss also dropped so so so slow.

You maybe miscalculated the loss of ArcMarginProduct.