Deep-Encoding-Pooling-Network-DEP- I used resnet50 as the baseline to train the MINC2500, and found that the accuracy was not up to 82% of the paper. Is this the result you ran out? Looking forward for your reply, thank you

I used resnet50 as the baseline to train the MINC2500, and found that the accuracy was not up to 82% of the paper. Is this the result you ran out? Looking forward for your reply, thank you

Open roseif opened this issue 4 years ago • 1 comments

Nov 16 '20 01:11 roseif

No, we used the SMOTE and Augmentor techniques to deal with an imbalanced dataset. Because of this, we surpassed the accuracy of our base paper.

Jul 01 '24 16:07 sumeetghumare4