HannH
HannH
My test shows it is about 0.3 mAP in Voc 2007 with EfficientDet-D2 too , I guess we need a pretrained model for EfficientDet to achieve mAP in paper.
> @HannH > emmmmmmm , why is just 0.3 map , did you train it enough epoch to convergence ? I test it in both trainset and testset, get the...
> @HannH Of course, welcome to release your reproduction code and post. I have released the code. But I'm not sure if align loss is writed in the right way....
change 1/8 to 1/16 in this line, you will get the correct grid number https://github.com/HannH/DMFN/blob/c12a42b2e0023bb52838afa4a14c48e4bfbe52f0/model/loss.py#L49