Adrian Bulat

@linkksng For compatibility reasons, in case someone prefers the old behavior probably its best to preserve the old model and logic too and simply mark it as deprecated. Depending on...

@moldach could you please install it from source, the issue should be fixed on master. I will push the changes to pip this weekend.

It doesn't like that you have installed the latest version: Current version: https://github.com/1adrianb/face-alignment/blob/2bcfcc6ea20d0104d48942737639827fd3d5e816/face_alignment/detection/sfd/detect.py#L15 Line present in the error: ```img = torch.from_numpy(img).to(device, dtype=torch.float32)```

While I haven't tested, it doesn't look like pytorch enforces explicitly 1.22 for numpy (https://github.com/pytorch/pytorch/blob/master/requirements.txt). If downgrading numpy to 1.21 doesn't work, you can either a) disable numba in the...

Hi Boris, Not really I am afraid. For simple use cases I was aligning the faces using a procrostus analysis, thought even for this I don't know of any existing...

For traced model you can find the links here: https://github.com/1adrianb/face-alignment/blob/c49ca6fef8ffa95a0ac7ce698e0b752ac91f6d42/face_alignment/api.py#L31 For normal weights you can find the links here: https://github.com/1adrianb/face-alignment/blob/ffc8bb499598c9076b28c3a2dc706b24c648b183/face_alignment/api.py#L39 and the corresponding model files here: https://github.com/1adrianb/face-alignment/blob/ffc8bb499598c9076b28c3a2dc706b24c648b183/face_alignment/models/fan.py

Sorry for delay. The network itself it relatively light and thin as such a high end GPU will typically be underutilized with it. To make better use of resources one...

@GuohongLi I have tested it on a 1080Ti and with a few optimisation I can get 20fps on a 640x480px video input ( fan+sfd )

@GuohongLi the smalles ones is 4x faster than the large one. The current model uploaded is "LARGE" (i.e. uses 4 stacks)

Currently, the landmarks follow the 68-points configuration used in the 300W dataset and challenge.