search

Qiming-Huang / ssformer

56 stars 8 forks source link

May I ask why the experimental results in the paper are so different from the experimental results in the github code base？ #7

Closed WangZhi-wz closed 1 year ago

WangZhi-wz commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Qiming-Huang commented 1 year ago

The result on Github is an old version (To avoid subsequent misleading, I removed the results displayed on GitHub), please refer to the final result depicted in the paper. Thanks.

sjc-lbj commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Can you fully reproduce the evaluation indicators in the paper？

WangZhi-wz commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Can you fully reproduce the evaluation indicators in the paper？

Yes, training on these two datasets individually basically achieves the results of the paper

sjc-lbj commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Can you fully reproduce the evaluation indicators in the paper？

Yes, training on these two datasets individually basically achieves the results of the paper

When I run testpy, only dice accuracy appears. May I ask how other indicators are repeated, such as iou?

sjc-lbj commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Can you fully reproduce the evaluation indicators in the paper？

Yes, training on these two datasets individually basically achieves the results of the paper Hello, may I ask how do you deal with the problem of multiple masks in the data science bowl 2018 set to recover the accuracy of the original paper

sjc-lbj commented 1 year ago

image image May I ask why the experimental results in the paper are so different from the experimental results in the github code base？

Can you fully reproduce the evaluation indicators in the paper？

Yes, training on these two datasets individually basically achieves the results of the paper Hello, may I ask how do you deal with the problem of multiple masks in the data science bowl 2018 set to recover the accuracy of the original paper