Skip to content

mem: free intermediate arrays during YoloX inference #483

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.

Sign up for GitHub

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

Open
KRRT7 wants to merge 2 commits into
base: main
Choose a base branch
from
Open

mem: free intermediate arrays during YoloX inference #483

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
4bfd7c4
mem: free intermediate arrays during YoloX inference
KRRT7 Mar 19, 2026
fe5b2b3
Add changelog entry and version bump to 1.5.3
KRRT7 Mar 19, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,3 +1,8 @@
## 1.5.3

### Enhancement
- Free intermediate arrays (`origin_img`, `img`, `ort_inputs`, `output`) at dead points during YoloX `image_processing()` to reduce peak memory during inference

## 1.5.2

### Fix
Expand Down
2 changes: 1 addition & 1 deletion unstructured_inference/__version__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1 +1 @@
__version__ = "1.5.2" # pragma: no cover
__version__ = "1.5.3" # pragma: no cover
3 changes: 3 additions & 0 deletions unstructured_inference/models/yolox.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -108,12 +108,15 @@ def image_processing(
input_shape = (1024, 768)
origin_img = np.array(image)
img, ratio = preprocess(origin_img, input_shape)
del origin_img # Free full-size image array before ONNX inference
session = self.model

ort_inputs = {session.get_inputs()[0].name: img[None, :, :, :]}
output = session.run(None, ort_inputs)
del img, ort_inputs # Free preprocessed inputs after inference
# TODO(benjamin): check for p6
predictions = demo_postprocess(output[0], input_shape, p6=False)[0]
del output

boxes = predictions[:, :4]
scores = predictions[:, 4:5] * predictions[:, 5:]
Expand Down