Open zhanghang1989 opened 3 years ago
Are these for object detection or instance segmentation? I'm looking for Mask Cascade R-CNN
specifically I'm looking for: https://s3.us-west-1.wasabisys.com/resnest/detectron/resnest200_detectron-02644020.pth
Detectron2 backbones are available at: https://drive.google.com/file/d/1dirZWyrfgl1w6Wrf4pYspvOKWS18198m/view?usp=sharing https://drive.google.com/file/d/1lGvb4Yk2uxQgzqHJn0JuOkfWuLIcSJv5/view?usp=sharing https://drive.google.com/file/d/1BhNxO8zdGxzAraEbtxKbuiPjzdLHhrJC/view?usp=sharing https://drive.google.com/file/d/16PbLixbWPpBj3HuhMc5fYuLqK3erIl9a/view?usp=sharing
thank you!
Hello, would it be possible to share a folder with all of the models we could download that were used in your encoding for semantic segmentation? Thanks!
Hello, would it be possible to share a folder with all of the models we could download that were used in your encoding for semantic segmentation? Thanks!
Thanks! There are quite a lot of models for semantic segmentation. I probably need to do it when I figure out a new way of releasing models. Do you have a specific model that you're interested in testing now?
Hello, would it be possible to share a folder with all of the models we could download that were used in your encoding for semantic segmentation? Thanks!
Thanks! There are quite a lot of models for semantic segmentation. I probably need to do it when I figure out a new way of releasing models. Do you have a specific model that you're interested in testing now?
Hi! @manningchuor and I are interested in testing EncNet_ResNet50s_ADE - thanks!
Hi @zhanghang1989, I had to disable resnest models in pytorch/hub for now https://github.com/pytorch/hub/pull/198 since the model weights are no longer available. A possible workaround is if less than 2GB, it’s recommended to attach it to a project release and use the url from the release. Let me know if you are ready to re-enable the test in pytorch hub. Thanks!
@ailzhang Thanks a lot for the suggestion! The GitHub release is a great solution.
I just made the changes at: https://github.com/zhanghang1989/ResNeSt/pull/156
The torch hub CI may be enabled. Thanks again.
@zhanghang1989 Could you please give me the certain model url used for ResNeSt-50-DCNv2 (ours) in detectron2?
The models weights can be found at https://github.com/zhanghang1989/ResNeSt/releases/tag/weights_step1
The one you are looking for is https://github.com/zhanghang1989/ResNeSt/releases/download/weights_step1/resnest50_detectron-255b5649.pth
Hi, I decided to follow this and try the semantic segmentation task. I executed the following code.
python scripts/prepare_ade20k.py
python test.py --dataset ADE20K --model-zoo EncNet_ResNet50s_ADE --eval
However, I got an error when downloading the pre-trained model.
The error message says that "resnet50s-a75c83cf.pth" is needed but is not locally available, so it is pulled from wasabi, and the URL is inaccessible due to restrictions.
I saved "resnest50-528c19ca.pth" from the above google drive URL to "/root/.encoding/models/" and ran test.py again, but got the same error.
So, could you please share "resnet50s-a75c83cf.pth" with us?
Due to hight downloading volume violate the Wasabi cloud rule, please manually download the models for now.
https://drive.google.com/file/d/10GXXl9ekgD3-npa7LiXZpQg6FI1bwpky/view?usp=sharing https://drive.google.com/file/d/1e-Z88a1c14Cwwn02CiBzQ2kFwO6OozHd/view?usp=sharing https://drive.google.com/file/d/1uSSvy4V7ALjousc7Tqy3tlGH7nLUijhG/view?usp=sharing https://drive.google.com/file/d/1dwMhRiuz-E7je-gK0mEFrIXp33MU2_FP/view?usp=sharing