hi @ZhangGongjie
Thanks you for your work and code.
I've been working on your code recently, but when I tried to reproduce your work, I found an obvious performance gap between your proposed dab_detr_imfa and the original dab_deformable_detr. Experiments are conducted on your uploaded codes. Is this normal?
And if possible, could you please upload your training log?
hi @ZhangGongjie
Thanks you for your work and code.
I've been working on your code recently, but when I tried to reproduce your work, I found an obvious performance gap between your proposed dab_detr_imfa and the original dab_deformable_detr. Experiments are conducted on your uploaded codes. Is this normal?
And if possible, could you please upload your training log?