sail-sg/volo

Issues

• Ablation study and official code problem

  #46 opened 2 years ago by xingshulicc
  0

• RuntimeError: The size of tensor a (28) must match the size of tensor b (14) at non-singleton dimension 2

  #39 opened 3 years ago by zhang-pan
  2

• Question about computational complexity formulation of Outlooker Attention

  #45 opened 2 years ago by ligeng0197
  0

• The Equ.5 and the operate `fold` in the paper do not seem to be consistent.

  #7 opened 3 years ago by lartpang
  6

• volo-d1 training without token label data

  #43 opened 3 years ago by michaeltrs
  1

• pre trained model file is broken.

  #44 opened 3 years ago by scotthuang1989
  1

• AttributeError: 'tuple' object has no attribute 'log_softmax'

  #33 opened 3 years ago by Snailgoo
  2

• size mismatch for pos_embed: copying a param with shape torch.Size([1, 14, 14, 768]) from checkpoint, the shape in current model is torch.Size([1, 14, 14, 512]).

  #38 opened 3 years ago by zhang-pan
  1

• Increasing GPU memory in every epoch when running volo-d2 without token labeling.

  #26 opened 3 years ago by Ree1s
  2

• When training own dataset, an error occurs when changing numberclasses to the corresponding category. If it is the default, it will report an error

  #42 opened 3 years ago by hx358031364
  0

• The code for Semantic Segmentation?

  #41 opened 3 years ago by HITerStudy
  0

• is the order of v_pj(x) result in algoritm1 correct ?

  #40 opened 3 years ago by TriLoo
  0

• Any plans for a tensorflow implementation

  #14 opened 3 years ago by king398
  2

• May you provide the FLOPs calculation script ?

  #37 opened 3 years ago by Sunting78
  0

• SystemError: returned NULL without setting an error

  #36 opened 3 years ago by JIAOJIAYUASD
  0

• run_error

  #35 opened 3 years ago by JIAOJIAYUASD
  1

• Could you share the training script for training volo_d5 ↑512?

  #34 opened 3 years ago by wangpichao
  0

• Will you release the codes for Semantic Segmentation?

  #32 opened 3 years ago by wangpichao
  0

• Balanced class weight

  #31 opened 3 years ago by javierrodenas
  1

• Finetune with 512 image size

  #30 opened 3 years ago by javierrodenas
  2

• Colab Notebook doesn't work & gives wrong results.

  #29 opened 3 years ago by debparth
  4

• TypeError: forward() takes 1 positional argument but 2 were given

  #22 opened 3 years ago by YHDASHEN
  2

• Some thoughts about volo

  #28 opened 3 years ago by cfzd
  0

• Halonets declare a SOTA in CVPR, any contribution to conference?

  #27 opened 3 years ago by inkzk
  0

• Compare to DynamicConv

  #5 opened 3 years ago by toodle
  3

• AttributeError: 'tuple' object has no attribute 'log_softmax'

  #25 opened 3 years ago by SamratSahoo
  2

• batch-size error

  #24 opened 3 years ago by k1anshul
  1

• Token Labeling

  #8 opened 3 years ago by monney
  9

• What's the difference between `x_cls` and `x_aux`?

  #23 opened 3 years ago by inkzk
  1

• intuition behind Outlook Attention Generation seems does not make sense

  #6 opened 3 years ago by toodle
  13

• how to interference images or videos？

  #20 opened 3 years ago by zxq309
  3

• outlook forward pass example

  #21 opened 3 years ago by ofTarradiddle
  1

• volo on object detection with detectron2?

  #3 opened 3 years ago by lucasjinreal
  5

• Could you please provide imagenet 22k pretrain ?

  #13 opened 3 years ago by zyxu1996
  1

• Confusion in class OutlookAttention moduel

  #12 opened 3 years ago by axhiao
  1

• colab

  #18 opened 3 years ago by AK391
  1

• password of token labeling 's Baidu Yun

  #19 opened 3 years ago by zhangyuge10
  0

• Main diference between "outlook attention" and "Involution".

  #1 opened 3 years ago by wuyongfa-genius
  7

• UnboundLocalError: local variable 'input' referenced before assignment

  #11 opened 3 years ago by zhang-pan
  1

• AttributeError: 'tuple' object has no attribute 'max'

  #10 opened 3 years ago by zhang-pan
  2

• There is a problem when loading the pretrained weights

  #2 opened 3 years ago by JonnesLin
  1

• Could you please release an ablation study that compares LV-VIT with / without outlook

  #9 opened 3 years ago by theFoxofSky
  4

• semantic segmentation

  #4 opened 3 years ago by ajtao
  3