Thanks for your inspiring work. On the polyp segmentation task, the ptotosam I ran on the kvasir dataset could not achieve the segmentation accuracy written in the paper.
Support index is [927] for several runs, and the test set is exactly given according to the link provided on github. The meanDice and meanIoU I run are 72.5 and 60.6 respectively, which are far from the results of 81.55 and 73.66 in the paper.
I didn't change any code and ran the experiment strictly according to the instruction, can you tell me which support image was selected for the results in the paper, and whether there is any step that I missed that prevented me from getting the results in the paper.
levayz
commented
1 week ago
Thanks for your inspiring work. On the polyp segmentation task, the ptotosam I ran on the kvasir dataset could not achieve the segmentation accuracy written in the paper. Support index is [927] for several runs, and the test set is exactly given according to the link provided on github. The meanDice and meanIoU I run are 72.5 and 60.6 respectively, which are far from the results of 81.55 and 73.66 in the paper. I didn't change any code and ran the experiment strictly according to the instruction, can you tell me which support image was selected for the results in the paper, and whether there is any step that I missed that prevented me from getting the results in the paper.