3DDFA icon indicating copy to clipboard operation
3DDFA copied to clipboard

The PyTorch improved version of TPAMI 2017 paper: Face Alignment in Full Pose Range: A 3D Total Solution.

Results 67 3DDFA issues
Sort by recently updated
recently updated
newest added

Can I use infrared images as input? @liguohao96

Hi, In **3DDFA**, inappropriate dependency versioning constraints can cause risks. Below are the dependencies and version constraints that the project is using ``` torch>=0.4.1 torchvision>=0.2.1 numpy>=1.15.4 scipy>=1.1.0 matplotlib==3.0.2 dlib==19.5.0 opencv-python>=3.4.3.18...

would you like to share the uncropped datasets?

Before proposing this issue, plz search the existed issues by your keywords first. **Describe the bug** **To Reproduce** **Expected behavior** **Screenshots** **Platform:** - macOS, Linux, or Windows **Additional context** Add...

How can I use the code to detect the pose of one face and then warp the other face to the same pose?

您好,我使用的是ubuntu18系统, python3 setup.py build_ext -i前的步骤都正常进行 这一步没有运行错误,但是显示 skipping 'mesh_core_cython.cpp' Cython extension (up-to-date) 结果这一步骤进行后,cython文件夹没有新的文件,请问如何解决,谢谢!

could you have python version about face profiling?

**Please detail your questions here : )** Sorry to bother, you said the strategy using VDC to finetune WPDC achieves the best result, I wonder the finetune operation in detail....

Could you plz provide the code that augment `300W-LP` to `train_aug_120x120`. I want to know how the `PoseParam` will change if an image is rotated or the bbox is perturbed....

I have some questions about the face profiling code. function **ImageMeshing** need **vertex_full** and **vertexm_full** as inputs. My question is : 1. what's the **vertex_full** and **vertexm_full** mean? and where...