Task Accuracy is too low on MFN dataset.

[a-runner]

I clone the main branch and download the datasets as detailed in the readme. The mIOU when reasoning on the MFN dataset differs too much from the paper, other datasets do not have this problem. The backbone is based on ftnet_resnext50_32x4d_filters_128_edge_3_blocks_2 and the file is under the rev-2 dir. Is there something wrong with my intermediate processing steps?

[shreyaskamathkm]

I am not really sure if the dataset has changed or if there is something wrong. Can you provide the steps you followed? I can take a look when I have sometime

[a-runner]

Thank you so much for your reply! I really appreciate your help.

• I first downloaded the dataset following the instructions in the readme.

• I then split the dataset according to the guidelines in MFNDataset_folderMap.py.

• For model inference, I used the testing commands from Codes/src/bash/Train_and_test_all.sh.

  • I used the weights from the path: "Codes/pretrained_models/ftnet_resnext50_32x4d_filters_128_edge_3_blocks_2/rev-2/MFN Dataset/ckpt/val_mIOU=0.4121.ckpt"
  • 

• The inference accuracy was normal on the SODA and SCUTSEG datasets. However, the inference accuracy was very low on the MFN dataset, and I'm unsure why this is the case.

[shreyaskamathkm]

Interesting, its been a while since i looked at this. I will get to it when I get time. The same code from the main branch was utilized to get the results.