microsoft / snr Goto Github PK

Do you have a plan to open person re-id task?

Some files are missing:
maskrcnn_benchmark.modeling.roi_heads.mask_head
maskrcnn_benchmark.modeling.rpn

Did I miss any installation step

hello, thank you so much for your sharing. But I'm not sure if the dual causality loss is used in detection, if not, why the dual causality loss is not used?

Thank you for your wonderful work. I noticed that you used entropy to distinguish task-relevant and task-irrelevant features, but entropy is usually not used in regression tasks. I wonder if SNR is applicable to regression tasks? Do you have any good suggestions for modifying it?

There seems to be dimension mismatch in the model.

Steps to produce:

model = resnet18_snr()
model(torch.randn(2, 3, 256, 128))

Error log:

~/Documents/Codes/re_id/deep-person-reid/torchreid/models/resnet_SNR.py in forward(self, x)
    276 
    277 
--> 278         x_4 = self.avgpool(x_4)
    279         x_4 = x_4.view(x_4.size(0), -1)
    280 

~/Documents/Codes/re_id/test/lib/python3.8/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
   1049         if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1050                 or _global_forward_hooks or _global_forward_pre_hooks):
-> 1051             return forward_call(*input, **kwargs)
   1052         # Do not call functions when jit is used
   1053         full_backward_hooks, non_full_backward_hooks = [], []

~/Documents/Codes/re_id/test/lib/python3.8/site-packages/torch/nn/modules/pooling.py in forward(self, input)
    613 
    614     def forward(self, input: Tensor) -> Tensor:
--> 615         return F.avg_pool2d(input, self.kernel_size, self.stride,
    616                             self.padding, self.ceil_mode, self.count_include_pad, self.divisor_override)
    617 

RuntimeError: Given input size: (512x8x4). Calculated output size: (512x2x-2). Output size is too small

1. I tried to reproduce the result for domain generalizable classification on PACS dataset but can't achieve the accuracy on both baseline(AGG) and SNR methods. Is there any trick or notes？What batch size shall I set？
2. May I ask which kind of activation map did you use in Fig5(b)? And what is the specified layer?

hi, thanks for your sharing. For the segmentation, why the dual causality loss is not used?

Thank you for sharing your code.

I want to execute your code but it needs the authority for accessing classification dataset in google drive.

How can i get the authority to the dataset?

Thank you.

There are open compliance tasks that need to be reviewed for your SNR repo.

Action required: 4 compliance tasks

To bring this repository to the standard required for 2021, we require administrators of this and all Microsoft GitHub repositories to complete a small set of tasks within the next 60 days. This is critical work to ensure the compliance and security of your microsoft GitHub organization.

Please take a few minutes to complete the tasks at: https://repos.opensource.microsoft.com/orgs/microsoft/repos/SNR/compliance

• The GitHub AE (GitHub inside Microsoft) migration survey has not been completed for this private repository
• No Service Tree mapping has been set for this repo. If this team does not use Service Tree, they can also opt-out of providing Service Tree data in the Compliance tab.
• No repository maintainers are set. The Open Source Maintainers are the decision-makers and actionable owners of the repository, irrespective of administrator permission grants on GitHub.
• Classification of the repository as production/non-production is missing in the Compliance tab.

You can close this work item once you have completed the compliance tasks, or it will automatically close within a day of taking action.

If you no longer need this repository, it might be quickest to delete the repo, too.

GitHub inside Microsoft program information

More information about GitHub inside Microsoft and the new GitHub AE product can be found at https://aka.ms/gim.

FYI: current admins at Microsoft include @lcl-2019