Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make VideoMAEImageProcessor much faster #28221

Closed
wants to merge 9 commits into from
7 changes: 6 additions & 1 deletion src/transformers/feature_extraction_utils.py
Original file line number Diff line number Diff line change
Expand Up @@ -135,8 +135,13 @@ def _get_is_as_tensor_fns(self, tensor_type: Optional[Union[str, TensorType]] =
raise ImportError("Unable to convert output to PyTorch tensors format, PyTorch is not installed.")
import torch # noqa

def recursive_ndarray_check(value):
if isinstance(value, (list, tuple)) and len(value) > 0:
return recursive_ndarray_check(value[0])
return isinstance(value, np.ndarray)

def as_tensor(value):
if isinstance(value, (list, tuple)) and len(value) > 0 and isinstance(value[0], np.ndarray):
if recursive_ndarray_check(value):
value = np.array(value)
return torch.tensor(value)

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -339,5 +339,7 @@ def preprocess(
for video in videos
]

data = {"pixel_values": videos}
# Speeds up tensor conversion - see: https://github.com/huggingface/transformers/pull/28221/files
data = {"pixel_values": np.asarray(videos) if return_tensors == "pt" else videos}
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this still needed?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Not really, although it doesn't hurt either. Should we remove it?

ikergarcia1996 marked this conversation as resolved.
Show resolved Hide resolved

return BatchFeature(data=data, tensor_type=return_tensors)
Loading