-
Hello, thanks for sharing the great work!
I found this piece of code in training/training_loop_recon_v1.py where you used **loss.gen_data_by_G_syn** to create synthetic images to train the Triplane…
-
and will it be problematic if you still use the RGB scale mean and std for normalizing the grey-scale image? which results in kinda high contrast dark blue faces before passing to eva-clip.
Thanks …
-
Hi, I'd like to know if there is a way to get the 68 landmarks coordinates in similarity aligned face image. I have the facial landmark coordinates in CSV file, but it's in original image, not in the …
-
### For?
ModelDoc
### What can't you do?
Currently when using gizmos for things like boxes, capsules and so on.
there isnt a nice way to perfectly align them to our models.
other than lots…
-
Post training both models have been fitted on all trainingsdata and new data (10th of April 2022 until 5th of june 2023).
This prediction gives an alarming amount of positives. Part of this presumabl…
-
The output img of `FaceAnalysis` is **BGR** image,but the codes in [IP-Adapter-FaceID](https://huggingface.co/h94/IP-Adapter-FaceID) and [hugging face docs](https://huggingface.co/docs/diffusers/v0.30…
-
Something like the following as described by Discord's drjth:
```py
def plane_from_face(face: Face, align=(Align.CENTER, Align.CENTER)):
u = (align[0].value - 1) / 2
v = (align[1].value - …
-
arcface_dst = np.array(
[[38.2946, 51.6963], [73.5318, 51.5014], [56.0252, 71.7366],
[41.5493, 92.3655], [70.7299, 92.2041]],
dtype=np.float32)
is provided in face_align.py for 5 ke…
-
This is a general issue about making it easier to style forms. Right now, we have an out of the box styling, and our general answer to the rest is "write custom CSS". But this is hard and error-prone.…
-
**Describe the solution you'd like**
I think there should be a few alignment tools for different objects. Currently as it sits, it's mostly just measuring distances and adjusting till it's lined up p…