PFLD-pytorch

PFLD-pytorch copied to clipboard

Bad quality of "eye" landmarks when eyes are closed.

1

Hello, great work there!!! I executed the camera.py file with your pre-trained model at "./checkpoint/snapshot/checkpoint.pth.tar". I realize that eye landmarks generated when eyes are closed are very bad (shown in...

HungVS

Can someone help me to understand "self.fc = nn.Linear(176, 196)" how 176 input feature has been choose

2

Can someone help me to understand `self.fc = nn.Linear(176, 196)` how 176 input feature has been choose in PFLDInference network.

sainisanjay

loss这里设计的是不是有问题？

4

weight_attribute = torch.sum(attributes_w_n.mul(mat_ratio), axis=1) loss里乘这一项导致没有特殊属性的人脸loss值为0，就是正常的人脸在训练里面的贡献为0，这样的设计正确么？还是我理解有误？

AiAiGhost

replace retinaface got wrong result

3

Hi, I simply replaced face detector, and the landmark generate is wrong: ``` [ 931.92762136 2339.20769691] [ 510.39959192 2297.27704525] [ 63.24823201 2284.217906 ] [ 494.01687384 2427.17235088] [ 744.55595613 1609.01716948] [...

lucasjinreal

Cannot find annotations

1

Hi! Nice work there. I could download he images from the link provided but Annotations are unavailable. Can You please share sample txt for annotations. I need to annotate my...

MuzammiFareed

想请问下为什么运行train时出现这种错误 应该如何修改

1

weight_angle = torch.sum(1 - torch.cos(angle - euler_angle_gt), axis=1) RuntimeError: The size of tensor a (3) must match the size of tensor b (0) at non-singleton dimension 1

daiguangq

脸离摄像头较近（脸占据照片的绝大部分）关键点会出现较大偏移，问题可能出现在哪里？如何解决？谢谢

houjinyan

Bumps [numpy](https://github.com/numpy/numpy) from 1.17.2 to 1.22.0. Release notes Sourced from numpy's releases. v1.22.0 NumPy 1.22.0 Release Notes NumPy 1.22.0 is a big release featuring the work of 153 contributors spread...

dependabot[bot]

使用onnx模型时得到的结果不相同/不一致 Result is different between using pytorch-model and onnx-model

4

python 3.7 + torch 1.8.1 ---中文--- 使用相同的输入，对pytorch模型 和 转换后的onnx模型进行测试，结果两个输出不一致，差别很大。 请问是什么原因，或者有人遇到相同问题吗？ 谢谢。 ---English--- I convert the pytorch model to onnx model using 'pytorch2onnx.py', then test the output of these two models....

zuozhen

version problem?

3

RuntimeError: ./checkpoint/snapshot/checkpoint.pth.tar is a zip archive (did you mean to use torch.jit.load()?)

DAVID-Hown