Skip to content

fix a NCHW pb converting bug in pad_rewriter #1063

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

Merged
merged 1 commit into from
Aug 21, 2020
+9 −3
Merged

fix a NCHW pb converting bug in pad_rewriter #1063

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
12 changes: 9 additions & 3 deletions tf2onnx/rewriter/conv2d_with_pad_rewriter.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,10 +43,16 @@ def rewrite_conv2d_with_pad(g, ops):
logger.debug("merge pad [%s] into conv [%s]", pad.name, conv.name)
paddings_val = np.array(paddings.get_tensor_value())
# can't pad on batch or channel dimensions
if np.any(paddings_val[0]) or np.any(paddings_val[3]):
continue
data_format = conv.get_attr("data_format").s.decode("utf-8")
if data_format == "NHWC":
if np.any(paddings_val[0]) or np.any(paddings_val[3]):
continue
paddings_val = paddings_val[1:3]
else:
if np.any(paddings_val[0]) or np.any(paddings_val[1]):
continue
paddings_val = paddings_val[2:4]

paddings_val = paddings_val[1:3]
paddings_val = paddings_val.transpose().flatten()
g.replace_input(conv, conv.input[0], pad.input[0])
# convert Conv2D
Expand Down