nkolot/GraphCMR

Issues

• How to get 3d joints from demo.py and visualize it

  #36 opened 5 years ago by maddyonline
  3

• Pretrained model

  #53 opened 2 years ago by lrh4653
  1

• The additional files could not be obtained.

  #52 opened 2 years ago by Accelerater-11
  0

• Provide code (not data) for Human3.6m + SMPL training

  #15 opened 5 years ago by benjiebob
  12

• wrong mesh volume

  #51 opened 2 years ago by friendly-code-bot
  0

• run demo.py

  #50 opened 2 years ago by lpx153623
  0

• preprocess datasets of h36m.py

  #48 opened 3 years ago by Hao-hui
  0

• Running ‘demo.py' can't get good results

  #41 opened 5 years ago by moondabaojian
  2

• Weak-perspective camera coordinates

  #1 opened 6 years ago by vladsterz
  7

• Asking for the weight of losses

  #44 opened 3 years ago by syguan96
  0

• how to retrain this model in a new dataset with the real SMPL model

  #45 opened 3 years ago by sunjc0306
  0

• The problem of camera parameter “sc”

  #42 opened 4 years ago by moondabaojian
  1

• Loading Resnet50 pretrained?

  #43 opened 4 years ago by thaithanhtuan
  0

• Praise from a newbie

  #38 opened 5 years ago by tinatiansjz
  1

• Why do you use different focal length for training and inference?

  #39 opened 5 years ago by krumo
  0

• About the SMPLParamRegressor

  #37 opened 5 years ago by MengXinChengXuYuan
  0

• Compute 'A', 'D', 'U' matrices

  #35 opened 5 years ago by moondabaojian
  6

• Regarding fully connected baseline

  #34 opened 5 years ago by mehtadushy
  4

• Preprocess of ground truth keypoints_3d on Human3.6M

  #31 opened 5 years ago by yuxwind
  1

• question when I read the code

  #32 opened 5 years ago by liwenssss
  1

• questions about "mesh_downsampling.npz"

  #33 opened 5 years ago by moondabaojian
  1

• How to train model with human3.6m datasets

  #30 opened 5 years ago by RanFeng
  1

• opendr?

  #25 opened 5 years ago by angechen
  3

• Multi-gpu training

  #26 opened 5 years ago by kevinlin311tw
  3

• Training on data of segmentation/DensePose prediction

  #28 opened 5 years ago by yuxwind
  1

• How to assign the correspoding vertices of smpl with respect to skeleton points?

  #29 opened 5 years ago by Maqingyang
  6

• openpose

  #27 opened 5 years ago by yuxwind
  0

• hyper-params differ from those in the paper?

  #11 opened 5 years ago by Maqingyang
  6

• Validation set accuracies (without human36 data)

  #18 opened 5 years ago by benjiebob
  3

• cpu occupacy is very high

  #20 opened 5 years ago by Maqingyang
  6

• About test performance

  #22 opened 5 years ago by zhaoyang10
  6

• demo img resolution

  #23 opened 5 years ago by lircsszz
  1

• Environment requirement

  #24 opened 5 years ago by CocoaPebble
  0

• There is a dangerous inplace operation in SMPL()

  #19 opened 5 years ago by Maqingyang
  3

• Train the graph adjacency matrix?

  #14 opened 5 years ago by sheoranhimansh
  1

• img_orig is the same as img

  #7 opened 5 years ago by yuxwind
  0

• What's the purpose of this processing on rotation matrix?

  #17 opened 5 years ago by Maqingyang
  4

• Training on Human3.6M provide Upside-down results

  #16 opened 5 years ago by Maqingyang
  4

• About loss_keypoints_3d

  #13 opened 5 years ago by SecretGuardian
  2

• Different J_regressor matrix for training and evaluation

  #12 opened 5 years ago by yuxwind
  3

• Generate 3D mesh for Human36M

  #9 opened 6 years ago by jeroenvb
  1

• checking the condition number for svd

  #10 opened 6 years ago by moshanATucsd
  2

• about the joints used

  #8 opened 6 years ago by weigq
  1

• Training parameters?

  #6 opened 6 years ago by gsygsy96
  3

• Could you provide MPII/annot files?

  #5 opened 6 years ago by yuxwind
  1

• where is preprocess_datasets.py?

  #4 opened 6 years ago by gsygsy96
  4

• Can you provide datasets/extras/h36m_valid_protocol1.npz?

  #3 opened 6 years ago by wei-mao-2019
  3

• About the training dataset of Human36M

  #2 opened 6 years ago by weigq
  6