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.

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

Fix: Resolves memory leak caused by using CRAFT detector with detect() or readtext(). #1278

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Fix: Resolves memory leak caused by using CRAFT detector with detect() or readtext(). #1278

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e992112
Resolves memory leak on detection.
daniellovera Jul 10, 2024
62773de
Adjusted to not call empty_cache() for non-Cuda
daniellovera Jul 16, 2024
bc53fc1
Correct cuda check
daniellovera Jul 16, 2024
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
8 changes: 8 additions & 0 deletions easyocr/detection.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,6 +45,9 @@ def test_net(canvas_size, mag_ratio, net, image, text_threshold, link_threshold,
with torch.no_grad():
Copy link

Choose a reason for hiding this comment

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

Suggested change
with torch.no_grad():
with torch.inference_mode():

Copy link
Author

Choose a reason for hiding this comment

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

Seems useful but perhaps should be a separate PR? I was focused on the GPU memory leak with this PR so not sure if they should be packaged together.

y, feature = net(x)

# remove x and feature from device, whether GPU or CPU
del x, feature

boxes_list, polys_list = [], []
for out in y:
# make score and link map
Expand All @@ -68,6 +71,11 @@ def test_net(canvas_size, mag_ratio, net, image, text_threshold, link_threshold,
polys[k] = boxes[k]
boxes_list.append(boxes)
polys_list.append(polys)

# remove y from device, whether GPU or CPU, and check if cuda was used before calling empty_cache() to clean up
del y
if device == 'cuda':
torch.cuda.empty_cache()

return boxes_list, polys_list

Expand Down