Tensor size mismatch getting predictions for a picture

Hi,

This is my code, modeled after the sample in description.

`import face_alignment`
`from skimage import io`

`model = face_alignment.FaceAlignment(landmarks_type= face_alignment.LandmarksType.THREE_D,device='cpu',flip_input=False)`

`input = io.imread('test.png')`
`preds = model.get_landmarks(input)`

Running it I get the error:

`Traceback (most recent call last):`

`File ~\anaconda3\Lib\site-packages\spyder_kernels\py3compat.py:356 in compat_exec
exec(code, globals, locals)`

`File facealignment.py:14
preds = model.get_landmarks(input)`

`File ~\anaconda3\Lib\site-packages\face_alignment\api.py:113 in get_landmarks
return self.get_landmarks_from_image(image_or_path, detected_faces, return_bboxes, return_landmark_score)`

` File ~\anaconda3\Lib\site-packages\torch\utils\_contextlib.py:116 in decorate_context
return func(*args, **kwargs)`

` File ~\anaconda3\Lib\site-packages\face_alignment\api.py:144 in get_landmarks_from_image
detected_faces = self.face_detector.detect_from_image(image.copy())`

`File ~\anaconda3\Lib\site-packages\face_alignment\detection\sfd\sfd_detector.py:45 in detect_from_image
bboxlist = detect(self.face_detector, image, device=self.device)[0]`

` File ~\anaconda3\Lib\site-packages\face_alignment\detection\sfd\detect.py:17 in detect
return batch_detect(net, img, device)`

` File ~\anaconda3\Lib\site-packages\face_alignment\detection\sfd\detect.py:33 in batch_detect
img_batch = img_batch - torch.tensor([104.0, 117.0, 123.0], device=device).view(1, 3, 1, 1)`

`RuntimeError: The size of tensor a (4) must match the size of tensor b (3) at non-singleton dimension 1`

I've tried different types of files, different models (dlib and default sfd, 2D and 3D) and it always results in the same. I'm running Python 3.11 and torch 2.5.1.

The error occurs in face_alignment/detection/sfd/detect.py at line 33 where a tensor size mismatch happens. The image likely has 4 channels (e.g., RGBA) but the model expects 3 (RGB). Start by checking the image shape in the debugger or by printing input.shape. Convert the image to RGB if needed. Look at the detect.py file to understand the preprocessing. Run the provided code with a known RGB image to verify the fix.

Rédigé par le modèle d'indexation à partir du texte de l'issue.