Hello, thank you very much for your work and code, I have a few questions that I am puzzled and hope you can help me.
I trained the small deit many times according to the parameters you gave, but the best results obtained on the CUB validation set are as follows:
Cls@1:0.782 Cls@5:0.945
Loc@1:0.626 Loc@5:0.764 Loc_gt:0.804
And the results given in the paper are:
Loc.Acc@1: 71.3 Loc.Acc@5: 83.8
Loc.Gt-Known: 87.7 Cls.Acc@1: 80.3 Cls.Acc@5: 94.8
How can I get the results in your paper? By increasing the number of iterations?
Thank you again, and look forward to your answers.
Hello, thank you very much for your work and code, I have a few questions that I am puzzled and hope you can help me.
I trained the small deit many times according to the parameters you gave, but the best results obtained on the CUB validation set are as follows:
Cls@1:0.782 Cls@5:0.945
Loc@1:0.626 Loc@5:0.764 Loc_gt:0.804
And the results given in the paper are:
Loc.Acc@1: 71.3 Loc.Acc@5: 83.8
Loc.Gt-Known: 87.7 Cls.Acc@1: 80.3 Cls.Acc@5: 94.8
How can I get the results in your paper? By increasing the number of iterations?
Thank you again, and look forward to your answers.