Question about evaluating the pre-trained model for cityscapes dataset

[Chenneyi]

Hello, when I evaluated the pre-trained model for cityscapes dataset, using the command: bash tools/dist_test.sh ./configs/Mobile_Seed/MS_tiny_cityscapes.py ckpt/MS_tiny_cityscapes.pth 1 --eval mIoU, the error occurred as follows: Have you ever come across this problem? How did you solve it?

[martin-liao]

Thank you for reporting this error! I have fixed the code for multi-gpu test, please pull the latest code and try again! tip: if you are testing with a single GPU, you can also change the --launcher pytorch to --launcher none for single-GPU testing.

[martin-liao]

The evaluation results on the Cityscapes dataset should be close to this:

[Chenneyi]

Thank you for reporting this error! I have fixed the code for multi-gpu test, please pull the latest code and try again! tip: if you are testing with a single GPU, you can also change the --launcher pytorch to --launcher none for single-GPU testing.

That works, thank you so much!!!!!!!!

[martin-liao]

Since the problem has been resolved, I will close the issue.