zcablii/LSKNet

Issues

• [Docs] FAIR1M 1.0

  #88 opened 5 days ago by Huyifan626
  1

• [Bug] RuntimeError: Default process group has not been initialized, please make sure to call init_process_group

  #87 opened 9 days ago by lixinru77
  3

• [Bug] ValueError: need at least one array to concatenate

  #47 opened 8 months ago by Xingjie-Han
  10

• [Bug] Very low speed problem in inference detector

  #85 opened a month ago by AliSharififar
  5

• Training time

  #86 opened a month ago by OE-jimvanoosten
  1

• DOTA1.0数据集

  #84 opened 2 months ago by Joey-He
  1

• 训练自己的数据集

  #83 opened 2 months ago by ocrhei
  10

• [Docs] 关于多尺度训练和测试

  #78 opened 3 months ago by LeoGeralt
  1

• [Docs] 请问单卡训练的话，需要改什么参数吗

  #80 opened 3 months ago by 921162820
  13

• 关于acc

  #82 opened 3 months ago by ocrhei
  3

• [Docs] 关于模型训练过程中的一些细节问题

  #81 opened 3 months ago by wap1024
  2

• FAIR1M pretrained weights

  #39 opened 9 months ago by tikitong
  3

• [Docs] 关于模型对比实验

  #79 opened 4 months ago by yuhua666
  1

• 训练LSKNet-T时map显示为0

  #77 opened 4 months ago by SMDR1412
  1

• [Bug]The mAP metric is too low.

  #75 opened 5 months ago by ZhenboZhao77
  4

• 使用单卡训练时报错

  #76 opened 5 months ago by adii-t
  2

• 空间选择和大核选择[Docs]

  #74 opened 5 months ago by ssfkfffddj
  6

• [Feature] 您好，读了好几遍您的LSK论文，其中最大的疑惑是怎么实现select的。 在代码中，就是将sig分成的两份，然后和不同尺度相乘，第一个sig和attn1相乘，第二个sig和attn2相乘，这是如何体现选择的？

  #73 opened 5 months ago by 23jisuper
  6

• [Docs]

  #72 opened 5 months ago by daihaoma
  2

• [Docs]model

  #71 opened 5 months ago by niupihhhh
  1

• Model deploy

  #70 opened 5 months ago by quxianjiuguo
  2

• KeyError: "OrientedRCNN: 'LSKNet is not in the models registry'"

  #69 opened 5 months ago by daihaoma
  2

• RuntimeError: Unable to find a valid cuDNN algorithm to run convolution

  #67 opened 6 months ago by lsjEric
  1

• [Feature] The pretrained weights on ImageNet are not avaliable now, could you provide them again?

  #68 opened 6 months ago by wuwuwuxxx
  1

• [Docs]Some questions about the code reproduction results of LSKNet_T in DOTA-1.0 dataset

  #51 opened 8 months ago by gbdjxgp
  8

• can not reproduce the result in log correctly

  #66 opened 6 months ago by CrazyBrick
  10

• Cannot reproduce lsknet_s_ss result on DOTA1.0

  #50 opened 8 months ago by Yiliud
  9

• which py files are necessary for the inference?

  #65 opened 6 months ago by jiofgnuo
  2

• [data augmentation]

  #63 opened 6 months ago by quxianjiuguo
  1

• 带有大的增长卷积核和增长膨胀率的深度可分离卷积

  #64 opened 6 months ago by ssfkfffddj
  2

• how can I prepare the Potsdam or Vaihingen datasets

  #62 opened 7 months ago by wxy16
  2

• [result]

  #61 opened 7 months ago by LW091
  10

• [split]

  #60 opened 7 months ago by LW091
  2

• [Feature] test

  #59 opened 7 months ago by LW091
  2

• [Docs] The feature dimension of 512 seems a bit small, perhaps there is a higher feature dimension

  #58 opened 7 months ago by fancy-girl
  1

• [Docs] can you give me some suggestions about the parameters?

  #57 opened 7 months ago by fancy-girl
  2

• [Docs] About HRSC2016 dataset

  #56 opened 8 months ago by LeiaJ520
  6

• Could you provide the pre-trained files? I would be very grateful.

  #55 opened 8 months ago by zly1213811
  4

• Default process group has not been initialized, please make sure to call init_process_group.

  #54 opened 8 months ago by fancy-girl
  2

• About where we can get the pretrained weight about the "LSKNet is pretrained on ImageNet for 100 or 300 epochs" in the article?

  #53 opened 8 months ago by xxxyyynnn
  2

• [Bug] Hi,can u tell me what tools do you use to draw the figure in your paper like Figure 4,thank you

  #52 opened 8 months ago by lzJune
  1

• Why I use lsk_s_ema_fpn_1x_dota_le90.py ad config and lsk_s_ema_fpn_1x_dota_le90_20230212-30ed4041.pth as checkpoint to test DOTAv1, the test mAP is only 81.3% rather than 81.85% reported on LSKNet official website?

  #49 opened 8 months ago by xxxyyynnn
  15

• load checkpoint from local path: /home/xyn02/LSKNet/configs/xynModel/lsk_s_ema_fpn_1x_dota_le90_20230212-30ed4041.pth The model and loaded state dict do not match exactly[Bug]

  #48 opened 8 months ago by xxxyyynnn
  4

• [Docs] Multi-scale train and test about DOTAv1.0

  #46 opened 8 months ago by LeiaJ520
  3

• How can i train and test on a new own data?

  #45 opened 8 months ago by kuaiqushangzixiba
  1

• About LSKmodule

  #41 opened 9 months ago by wawjwyy123
  10

• [Bug] 你好，我试着安装mmrotate的框架，在测试demo的时候运行image_demo.py，只有运行设备为CPU的时候好使，换其他任何一块显卡最终输出的都是原图，照着您的配置我觉得没啥问题，请帮我看一下哪出了问题？谢谢

  #44 opened 8 months ago by lzJune
  5

• [Bug] TypeError: FormatCode() got an unexpected keyword argument 'verify'

  #42 opened 9 months ago by Pluviophile-sy
  2

• [shceduler]

  #43 opened 9 months ago by spurs30
  3

• Questions about image split

  #40 opened 9 months ago by yyq0828
  1