If I replace the landmark detection model with 68-point version from your Google Drive link, I get error:
Traceback (most recent call last):
File "/home/gulingna/OpenFace3/openface3_landmark.py", line 9, in
landmark_detector = LandmarkDetector(model_path='./openface3_models/Landmark_68.pkl')
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/openface/landmark_detection.py", line 27, in init
self.alignment = Alignment(
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/openface/STAR/demo.py", line 117, in init
net.load_state_dict(checkpoint["net"])
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/torch/nn/modules/module.py", line 2637, in load_state_dict
raise RuntimeError(
RuntimeError: Error(s) in loading state_dict for StackedHGNetV1:
size mismatch for out_heatmaps.0.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).
size mismatch for out_heatmaps.0.conv.bias: copying a param with shape torch.Size([68]) from checkpoint, the shape in current model is torch.Size([98]).
size mismatch for out_heatmaps.1.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).
size mismatch for out_heatmaps.1.conv.bias: copying a param with shape torch.Size([68]) from checkpoint, the shape in current model is torch.Size([98]).
size mismatch for out_heatmaps.2.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).
If I replace the landmark detection model with 68-point version from your Google Drive link, I get error:
Traceback (most recent call last):
File "/home/gulingna/OpenFace3/openface3_landmark.py", line 9, in
landmark_detector = LandmarkDetector(model_path='./openface3_models/Landmark_68.pkl')
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/openface/landmark_detection.py", line 27, in init
self.alignment = Alignment(
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/openface/STAR/demo.py", line 117, in init
net.load_state_dict(checkpoint["net"])
File "/home/gulingna/miniconda3/envs/py310/lib/python3.10/site-packages/torch/nn/modules/module.py", line 2637, in load_state_dict
raise RuntimeError(
RuntimeError: Error(s) in loading state_dict for StackedHGNetV1:
size mismatch for out_heatmaps.0.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).
size mismatch for out_heatmaps.0.conv.bias: copying a param with shape torch.Size([68]) from checkpoint, the shape in current model is torch.Size([98]).
size mismatch for out_heatmaps.1.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).
size mismatch for out_heatmaps.1.conv.bias: copying a param with shape torch.Size([68]) from checkpoint, the shape in current model is torch.Size([98]).
size mismatch for out_heatmaps.2.conv.weight: copying a param with shape torch.Size([68, 256, 1, 1]) from checkpoint, the shape in current model is torch.Size([98, 256, 1, 1]).