ProDA
ProDA copied to clipboard
microsoft
Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation (CVPR 2021)
Results
22
ProDA issues
Sort by
recently updated
recently updated
newest added
trafficstars
Which pictures of Cityscapes dataset are use to make _qualitative results of semantic segmentation_?
1
Good job!! I would like to ask which pictures of Cityscapes dataset are use to make _qualitative results of semantic segmentation_, 500 val images? thanks!
why the softmax output "teacher" will equal to 250 (ignore index) here? https://github.com/microsoft/ProDA/blob/9ba80c7dbbd23ba1a126e3f4003a72f27d121a1f/models/adaptation_modelv2.py#L316 Is it a bug? Is it should be "threshold_arg"?
