Daniil-Osokin/lightweight-human-pose-estimation.pytorch

Issues

• How do I set the training parameters?

  #330 opened 3 months ago by amYaniii
  3

• C++ demo

  #327 opened 8 months ago by JuanF2019
  1

• Lines

  #326 opened 8 months ago by ywqly
  2

• python demo

  #329 opened 8 months ago by ZHAJIANGMIAN2002
  2

• Train on custom dataset question

  #324 opened a year ago by cyhsien
  2

• heat maps vs. confidence maps

  #322 opened a year ago by ywqly
  4

• change the size of the input of the network

  #321 opened a year ago by WalterSenpai
  2

• the model not giving straight skeleton

  #320 opened a year ago by Nimaro76
  4

• MobilenetV2

  #319 opened a year ago by ywqly
  2

• Evaluation metrics

  #318 opened a year ago by sathishram12
  2

• fps

  #314 opened 2 years ago by ywqly
  4

• Why does the ap drop when Dilated MobileNet v2 is used?

  #317 opened 2 years ago by xin4431
  2

• I also encountered this ：problem:AssertionError: Results do not correspond to current coco set！

  #316 opened 2 years ago by xin4431
  1

• Problem of group_keypoints

  #306 opened 2 years ago by yoletPig
  10

• Is it normal that the loss function does not converge?

  #315 opened 2 years ago by xin4431
  1

• AssertionError: Results do not correspond to current coco set

  #313 opened 2 years ago by ywqly
  9

• Errors encountered during validation. Nothing is written in the detections.json file.

  #304 opened 2 years ago by YOOHYOJEONG
  11

• The problem of estimating keypoint of wrist

  #309 opened 2 years ago by BigMouthFlower
  3

• some problems during the training process

  #312 opened 2 years ago by WidCoder
  3

• ap

  #311 opened 2 years ago by ywqly
  6

• How to solve the occlusion problem during human tracking

  #310 opened 2 years ago by liudadan
  2

• Confident pose larger 1

  #308 opened 2 years ago by codewithleo1103
  2

• Question about the BODY_PARTS_PAF_IDS setting

  #291 opened 2 years ago by sulei1998
  1

• Is mask a background image? I output the mask. It seems that the value of most positions is 1. Which positions should be 0?

  #305 opened 2 years ago by liudadan
  2

• The size of pictures during training is [3, 368, 368], and the size of pictures during verification is not uniform. Why unify the picture size during training. Why not unify the image size when verifying.

  #302 opened 2 years ago by liudadan
  2

• Why scale the mask size? Does this damage the label information and cause the model accuracy to drop?

  #301 opened 2 years ago by liudadan
  2

• Why should a mask value be stored with multiple variables? Can't we use a variable to store the mask?

  #300 opened 2 years ago by liudadan
  2

• Help me!! object of type <class 'numpy.float64'> cannot be safely interpreted as an integer.

  #303 opened 2 years ago by YOOHYOJEONG
  3

• How to change the number of downsample?

  #299 opened 2 years ago by sulei1998
  2

• Where is CocoTrainDataSet (Dataset) called__ getitem__ (self, idx) function

  #298 opened 2 years ago by liudadan
  2

• OSError: Video 0 cannot be opened

  #297 opened 2 years ago by leonhard-yu-zhang
  2

• Why do you re create the mask without using the label of the heatmap and the label of the paf as the mask

  #296 opened 2 years ago by liudadan
  2

• Why model uses BN layer

  #295 opened 2 years ago by liudadan
  2

• train question

  #294 opened 2 years ago by sulei1998
  4

• Training with my own datasets

  #293 opened 2 years ago by ted1986122
  2

• The coco dataset contains 17 key points Why can your model identify 18 key points.

  #292 opened 2 years ago by liudadan
  2

• keypoint

  #288 opened 2 years ago by 15154278491
  3

• Nan Loss

  #290 opened 2 years ago by sulei1998
  1

• demo question

  #289 opened 2 years ago by sulei1998
  4

• Modifying val.py for custom dataset

  #287 opened 2 years ago by dtkutzke
  2

• key point coordinates

  #286 opened 2 years ago by ywqly
  2

• The processing time of extract_keypoints

  #284 opened 2 years ago by shaohua-pan
  6

• Can I use this in my research work?

  #285 opened 2 years ago by littleJeffrey
  4

• why square video more slower than 16:9 video

  #282 opened 2 years ago by reflelia
  2

• mobilenet_ sgd_ 68.848.pth.tar is the pre training model of Imagenet

  #283 opened 2 years ago by Deep-learning999
  2

• Can I estimate only single person?

  #281 opened 2 years ago by reflelia
  7

• 报“WARNING”

  #280 opened 2 years ago by xieyifeng
  3

• Custom Dataset Annotation format

  #278 opened 2 years ago by JJLimmm
  5

• paf_masks

  #279 opened 2 years ago by XiaoJiNu
  3

• Fine Tuning query

  #277 opened 2 years ago by leftbackn3
  2