github-luffy/PFLD_68points_Pytorch

Issues

• LICENSE

  #55 opened 10 months ago by mkurczew
  0

• 关于损失函数的问题

  #53 opened 2 years ago by Jackshi1997
  1

• 请问你这边提供的预训练模型是使用的什么数据集训练的？

  #38 opened 3 years ago by brisyramshere
  1

• 98 点 转 68 点的眼睛的关键点没对应上

  #54 opened 2 years ago by udbhssnhwgf
  0

• 好像给dataloader设置num_worker就会出bug？

  #52 opened 2 years ago by biepenghaomie
  0

• 没有依赖包版本文件

  #49 opened 2 years ago by young169
  2

• model.load_state_dict(torch.load('efficientNet-b0.pth'))

  #30 opened 3 years ago by onlinehuazai
  8

• 关于用自己制作的人脸数据集应该如何预处理的问题？

  #51 opened 2 years ago by HGS-1998
  0

• 欧拉角这部分怎么理解的呢

  #50 opened 2 years ago by txw-github
  1

• 关于coco里的关键点

  #48 opened 2 years ago by PureHing
  3

• 脸离摄像头较近关键点出现较大偏移

  #45 opened 2 years ago by houjinyan
  1

• 有关3D人脸坐标的问题

  #42 opened 3 years ago by leizhu1989
  1

• Does model also provides each landmark confidence score

  #47 opened 2 years ago by sainisanjay
  3

• Loss saturate after 80 Epoch

  #46 opened 2 years ago by sainisanjay
  1

• 眼部单独训练

  #33 opened 3 years ago by chuck0518
  45

• 正常最终的损失值大概为多少？

  #34 opened 3 years ago by huangzhenjie
  7

• test代码的算error分母好像不对

  #44 opened 3 years ago by Incandescenced1
  1

• 关于使用300W数据集进行训练的问题

  #41 opened 3 years ago by axhionning
  3

• 还想没有300W的困难数据的txt文件（ibug的130多张图的）

  #43 opened 3 years ago by MachineLearning11
  0

• 训练多少个epoch后loss能正常收敛

  #40 opened 3 years ago by leizhu1989
  8

• 读取数据出错

  #39 opened 3 years ago by leizhu1989
  0

• train_model.py中的一点疑问

  #37 opened 3 years ago by 7aughing
  2

• 不同img_cropped区域得到的关键点效果差距较大

  #28 opened 3 years ago by 7aughing
  9

• 有没有使用PFLD做头部姿态估计的代码，而不是使用dlib

  #36 opened 3 years ago by MachineLearning11
  0

• 300w数据集的百度云密码是多少

  #35 opened 3 years ago by MachineLearning11
  2

• loss or NME computing for convergence

  #26 opened 4 years ago by epoc88
  5

• 可视化标注数据

  #32 opened 3 years ago by kaikaizhu
  8

• model = torch.load('efficientNet-b0.pth')

  #31 opened 3 years ago by onlinehuazai
  1

• Environment setup?

  #29 opened 3 years ago by ithmz
  0

• 关于为什么要用Auxiliary net的问题

  #24 opened 4 years ago by wx19941204
  8

• 检测速度

  #18 opened 4 years ago by txy1680
  7

• 300W Dataset baiduyun 验证码是?

  #27 opened 4 years ago by hushunda
  1

• What's the final loss ?

  #13 opened 4 years ago by PigletOS
  2

• torch.nn.modules.module.ModuleAttributeError: 'MobileNetV2' object has no attribute 'module'

  #25 opened 4 years ago by epoc88
  0

• 百度云提取码可以提供下么？

  #23 opened 4 years ago by mtllyb
  1

• NO config information of MobilenetV2

  #22 opened 4 years ago by wx19941204
  3

• 模型转换部署

  #21 opened 4 years ago by jiangjiliang
  0

• AuxiliaryNet的优化问题

  #20 opened 4 years ago by Qidian213
  2

• 请问ResNest50模型结果是如何训练得到的呀？

  #19 opened 4 years ago by my-yy
  2

• 请问有resnet50检测人脸关键点的代码吗？

  #16 opened 4 years ago by zhouqianyu0918
  1

• 300W的setpreparation部分

  #17 opened 4 years ago by zhou-th
  3

• CPU上的测试时间？

  #9 opened 4 years ago by geoffzhang
  1

• 代码的数据处理中没看到crop部分

  #10 opened 4 years ago by ganqi91
  1

• 关于数据集处理

  #12 opened 4 years ago by Brystone
  5

• "getNewFacialLandmarksFromFacePP.py" is incomplete

  #15 opened 4 years ago by xieshenru
  2

• the download address of Baidu cloud for300W Dataset

  #14 opened 4 years ago by xieshenru
  5

• 代码loss中没有考虑类不平衡问题的属性权重？

  #8 opened 4 years ago by YjDai
  7

• 模型

  #11 opened 4 years ago by billionfish
  3

• 您好，您这个代码不是使用的PFLD的PFLDLoss而是wing_loss？所以这个和PFLD方法的关系是什么？

  #7 opened 4 years ago by liangsen-zju
  1

• ModuleNotFoundError: No module named 'tensorflow'

  #6 opened 4 years ago by changfali
  0