Skip to content

Add Resize field; Fix Clip and Upscale2D layer #123

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

Add Resize field; Fix Clip and Upscale2D layer #123

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
915d831
add min and max params for Clip layer
MaximKuklin Jun 2, 2021
8f8a6ec
scale params are placed in node.input[2], not in node.input[1]
MaximKuklin Jun 2, 2021
48de576
Add Resize key to available layers
MaximKuklin Jun 2, 2021
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
1 change: 1 addition & 0 deletions onnx2keras/layers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,4 +70,5 @@
'Pad': convert_padding,
'Flatten': convert_flatten,
'Upsample': convert_upsample,
'Resize': convert_upsample
}
5 changes: 5 additions & 0 deletions onnx2keras/operation_layers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -28,6 +28,11 @@ def convert_clip(node, params, layers, lambda_func, node_name, keras_name):

input_0 = ensure_tf_type(layers[node.input[0]], name="%s_const" % keras_name)

if 'min' not in params or 'max' not in params:
node_input_data = [layers[attr] for attr in node.input[1:]]
max_attr, min_attr = max(node_input_data), min(node_input_data)
Copy link

Choose a reason for hiding this comment

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

If min only omitted - max will be replaced. Is ts correct behavior?

params["max"], params["min"] = max_attr, min_attr

if params['min'] == 0:
logger.debug("Using ReLU({0}) instead of clip".format(params['max']))
layer = keras.layers.ReLU(max_value=params['max'], name=keras_name)
Expand Down
3 changes: 2 additions & 1 deletion onnx2keras/upsampling_layers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,7 +25,8 @@ def convert_upsample(node, params, layers, lambda_func, node_name, keras_name):
else:
# for opset version - 9+
# Upsample since opset version 9 uses input[1] as 'scales' instead of attributes.
scale = np.uint8(layers[node.input[1]][-2:])
#
scale = np.uint8(layers[node.input[2]][-2:])

if params['mode'].decode('utf-8') != 'nearest':
logger.error('Cannot convert non-nearest upsampling.')
Expand Down