microsoft/CSWin-Transformer

Issues

• GPU and Precision Information for Cascade MaskRCNN Throughput Results

  #50 opened a month ago by igm503
  0

• timm version

  #49 opened 9 months ago by haveasip-and-code
  1

• The weight size of pretrained model Cswin-L is not the same with the script?

  #34 opened 3 years ago by Sunting78
  1

• The num_heads of CSWin-B in the paper and code are not the same

  #18 opened 3 years ago by 282857341
  1

• Experiment setting for semantic segmentation

  #13 opened 3 years ago by AnukritiSinghh
  2

• I tested several segmentation data, and I doubt your performance

  #20 opened 3 years ago by zhouyi-git
  5

• The problem is shown in the figure

  #5 opened 3 years ago by rui-cf
  4

• The results of downstream task by my realization are poor

  #17 opened 3 years ago by Sunting78
  3

• It's was hard to fine-tuning on other dataset.

  #14 opened 3 years ago by BboyHanat
  5

• What is optimized drop_path_rate for cSwinLarge384?

  #47 opened 2 years ago by osivaz61
  0

• Welcome update to OpenMMLab 2.0

  #46 opened 2 years ago by vansin
  0

• when i use the cswin as backbone, but the input resolution is different with 224, can i change the width from 1, 2, 7, 7 to 1, 2, 8, 8?

  #45 opened 2 years ago by zxyyxzz
  0

• How do you produce Table 9 (ablation on different attention mecahnisms) in the paper?

  #44 opened 2 years ago by rayleizhu
  0

• about dataset cifar100

  #43 opened 2 years ago by he-mixedSon
  0

• Function of drop_rate, attn_drop_rate and drop_path_rate——drop_rate, attn_drop_rate and drop_path_rate应该设置多少，模型能提高map喃

  #37 opened 2 years ago by LUO77123
  4

• about the setting of --use_chk

  #42 opened 2 years ago by go-ahead-maker
  0

• The split size on ADE20K

  #10 opened 3 years ago by LTnanana
  4

• Input image normalization parameters for semantic segmentation.

  #41 opened 2 years ago by NikAleksFed
  0

• The label of Imagenet 22k

  #29 opened 3 years ago by sweetorangezhuyan
  3

• In the CSwin-Transformer segmentation page, dose Fine-tuning mean pre-trained model?

  #40 opened 2 years ago by aihcyllop
  0

• Can CSWin run on Windows10 systems?

  #33 opened 3 years ago by aihcyllop
  1

• Using transfer to train over food101

  #39 opened 2 years ago by andreynz691
  0

• Inference Strategy

  #38 opened 2 years ago by rayleizhu
  0

• Hi, very glad to see this new version of Swin-Trans. Could I have a question about using mixed-precision training

  #35 opened 3 years ago by SudongCAI
  2

• when you release the code for object detection?

  #36 opened 3 years ago by jiangzhengkai
  1

• Are Training and Fintuning log available？

  #32 opened 3 years ago by zws98
  0

• Finetune settings for CSWin-B with 224x224 image size

  #31 opened 3 years ago by zws98
  0

• the memory is nearly twice that of swin transformer

  #23 opened 3 years ago by cjt222
  1

• 'CSWin is not in the backbone registry'

  #30 opened 3 years ago by ntuyt
  0

• CSWin-L for sota performance ade20k

  #28 opened 3 years ago by YellowPig-zp
  0

• error

  #27 opened 3 years ago by nhathoang0110
  1

• imagenet-22k pretrained cswin-B

  #26 opened 3 years ago by ydhongHIT
  0

• about the test result on imagenet-2012

  #24 opened 3 years ago by rentainhe
  1

• Could you release COCO object detection training codes and configs?

  #25 opened 3 years ago by JeremieMelo
  0

• Hi, this is a nice job, and I would like to ask whether CSWin-Tranformer has a corresponding 3D version? Just like Swin Transformer has a corresponding Video Swin Transformer. : )

  #22 opened 3 years ago by BossZard
  0

• About the patches_resolution of the segmentation model

  #21 opened 3 years ago by danczs
  2

• How to finetune on ADE20K?

  #3 opened 3 years ago by lihua7351
  1

• CSwin Code for Segmentation with MMSegmentation

  #15 opened 3 years ago by WalBouss
  1

• semantic segmentation

  #16 opened 3 years ago by businiaoo
  1

• when to release the code?

  #2 opened 3 years ago by zimenglan-sysu-512
  4

• 我测试了几个分割数据，有点怀疑你的性能 I tested several segmentation data, and I doubt your performance

  #19 opened 3 years ago by zhouyi-git
  0

• CSWin significantly slower than Swin?

  #11 opened 3 years ago by ErenBalatkan
  2

• why gpu memory of cswin-tiny is twice than swin-tiny?

  #12 opened 3 years ago by UcanSee
  0

• Pretrained settings for object detection

  #8 opened 3 years ago by youngwanLEE
  2

• Inconsistent hyper-parameter configuration between the code and arxiv report

  #9 opened 3 years ago by tonysy
  1

• Error about building 384 models

  #6 opened 3 years ago by TingquanGao
  2

• how to set param drop-path for base and large model?

  #4 opened 3 years ago by zimenglan-sysu-512
  2