onnx2tf 1.27.1__tar.gz → 1.27.3__tar.gz

{onnx2tf-1.27.1/onnx2tf.egg-info → onnx2tf-1.27.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: onnx2tf
-Version: 1.27.1
+Version: 1.27.3
 Summary: Self-Created Tools to convert ONNX files (NCHW) to TensorFlow/TFLite/Keras format (NHWC). The purpose of this tool is to solve the massive Transpose extrapolation problem in onnx-tensorflow (onnx-tf).
 Home-page: https://github.com/PINTO0309/onnx2tf
 Author: Katsuya Hyodo
@@ -334,7 +334,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  ghcr.io/pinto0309/onnx2tf:1.27.1
+  ghcr.io/pinto0309/onnx2tf:1.27.3
 
   or
 
@@ -342,7 +342,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  docker.io/pinto0309/onnx2tf:1.27.1
+  docker.io/pinto0309/onnx2tf:1.27.3
 
   or
 

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/README.md

@@ -309,7 +309,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  ghcr.io/pinto0309/onnx2tf:1.27.1
+  ghcr.io/pinto0309/onnx2tf:1.27.3
 
   or
 
@@ -317,7 +317,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  docker.io/pinto0309/onnx2tf:1.27.1
+  docker.io/pinto0309/onnx2tf:1.27.3
 
   or
 

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/onnx2tf/__init__.py

@@ -1,3 +1,3 @@
 from onnx2tf.onnx2tf import convert, main
 
-__version__ = '1.27.1'
+__version__ = '1.27.3'

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/onnx2tf/onnx2tf.py

@@ -1259,11 +1259,6 @@ def convert(
                 if len(msg_list) > 0:
                     for s in msg_list:
                         if 'Unable to serialize VariableSpec' in s:
-                            warn(
-                                f'This model contains GroupConvolution and is automatically optimized for TFLite, ' +
-                                f'but is not output because h5 does not support GroupConvolution. ' +
-                                f'If h5 is needed, specify --disable_group_convolution to retransform the model.'
-                            )
                             break
                 else:
                     error(e)
@@ -1285,11 +1280,6 @@ def convert(
                 if len(msg_list) > 0:
                     for s in msg_list:
                         if 'Unable to serialize VariableSpec' in s:
-                            warn(
-                                f'This model contains GroupConvolution and is automatically optimized for TFLite, ' +
-                                f'but is not output because keras_v3 does not support GroupConvolution. ' +
-                                f'If keras_v3 is needed, specify --disable_group_convolution to retransform the model.'
-                            )
                             break
                 else:
                     error(e)
@@ -1404,10 +1394,7 @@ def convert(
                 )
                 info(Color.GREEN(f'TFv1 .pb output complete!'))
             except KeyError as e:
-                warn(
-                    f'Probably due to GroupConvolution, saved_model could not be generated successfully, ' +
-                    f'so onnx2tf skip the output of TensorFlow v1 pb.'
-                )
+                pass
             except Exception as e:
                 error(e)
                 import traceback

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/onnx2tf/ops/Conv.py

@@ -468,12 +468,6 @@ def make_node(
                 error_check_tf_op_type = 'conv_bias'
 
             else:
-                if kernel_size in (1, 2, 3) and not disable_group_convolution:
-                    warn(
-                        f'This model contains GroupConvolution and is automatically optimized for TFLite, ' +
-                        f'but is not output because saved_model does not support GroupConvolution. ' +
-                        f'If saved_model is needed, specify --disable_group_convolution to retransform the model.'
-                    )
                 # GroupedConvolution - Conv1D, Conv2D, Conv3D - Bias Add
                 if kernel_size == 1 and not disable_group_convolution:
                     tf_layers_dict[graph_node_output.name]['tf_node'] = \
@@ -571,12 +565,6 @@ def make_node(
                 error_check_tf_op_type = 'conv_nobias'
 
             else:
-                if kernel_size in (1, 2, 3) and not disable_group_convolution:
-                    warn(
-                        f'This model contains GroupConvolution and is automatically optimized for TFLite, ' +
-                        f'but is not output because saved_model does not support GroupConvolution. ' +
-                        f'If saved_model is needed, specify --disable_group_convolution to retransform the model.'
-                    )
                 # GroupedConvolution - Conv1D, Conv2D, Conv3D - No Bias
                 if kernel_size == 1 and not disable_group_convolution:
                     tf_layers_dict[graph_node_output.name]['tf_node'] = \

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/onnx2tf/ops/ConvInteger.py

@@ -417,12 +417,6 @@ def make_node(
             error_check_tf_op_type = 'conv_nobias'
 
         else:
-            if kernel_size in (1, 2, 3) and not disable_group_convolution:
-                warn(
-                    f'This model contains GroupConvolution and is automatically optimized for TFLite, ' +
-                    f'but is not output because saved_model does not support GroupConvolution. ' +
-                    f'If saved_model is needed, specify --disable_group_convolution to retransform the model.'
-                )
             # GroupedConvolution - Conv1D, Conv2D, Conv3D - No Bias
             if kernel_size == 1 and not disable_group_convolution:
                 tf_layers_dict[graph_node_output.name]['tf_node'] = \

{onnx2tf-1.27.1 → onnx2tf-1.27.3}/onnx2tf/ops/Split.py

@@ -158,15 +158,56 @@ def make_node(
                 if idx == axis:
                     end_.append(split_idx + 1)
                 elif input_tensor_shape[idx] is None:
-                    end_.append(-1)
+                    end_.append(0)
                 else:
                     end_.append(input_tensor_shape[idx])
 
+            begin_mask_ = np.sum([2**idx if idx != axis else 0 for idx in range(input_tensor_rank)])
+            end_mask_ = np.sum([2**idx if idx != axis else 0 for idx in range(input_tensor_rank)])
+
             splited_tensors.append(
                 tf.strided_slice(
                     input_=input_tensor,
                     begin=begin_,
                     end=end_,
+                    begin_mask=begin_mask_,
+                    end_mask=end_mask_,
+                )
+            )
+    elif isinstance(split, np.ndarray) \
+        and len(list(split)) > 1 \
+        and np.prod(split) != 1 \
+        and np.all(split == split[0]) \
+        and isinstance(input_tensor_shape[axis], int) \
+        and input_tensor_shape[axis] == np.sum(split):
+        # strided_slice - Slice everything in same size
+        # Suppression of FlexSplitV generation
+        # https://github.com/PINTO0309/onnx2tf/issues/751
+        splited_tensors = []
+        split_size = split[0]
+        for split_idx in range(len(list(split))):
+            begin_ = [
+                split_size * split_idx if idx == axis else 0 for idx in range(input_tensor_rank)
+            ]
+            end_ = []
+            for idx in range(input_tensor_rank):
+                if idx == axis:
+                    end_.append(split_size * split_idx + split_size)
+                elif input_tensor_shape[idx] is None:
+                    end_.append(0)
+                else:
+                    end_.append(input_tensor_shape[idx])
+
+            begin_mask_ = np.sum([2**idx if idx != axis else 0 for idx in range(input_tensor_rank)])
+            end_mask_ = np.sum([2**idx if idx != axis else 0 for idx in range(input_tensor_rank)])
+
+            splited_tensors.append(
+                tf.strided_slice(
+                    input_=input_tensor,
+                    begin=begin_,
+                    end=end_,
+                    begin_mask=begin_mask_,
+                    end_mask=end_mask_,
                 )
             )
     elif isinstance(split, np.ndarray) \

{onnx2tf-1.27.1 → onnx2tf-1.27.3/onnx2tf.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: onnx2tf
-Version: 1.27.1
+Version: 1.27.3
 Summary: Self-Created Tools to convert ONNX files (NCHW) to TensorFlow/TFLite/Keras format (NHWC). The purpose of this tool is to solve the massive Transpose extrapolation problem in onnx-tensorflow (onnx-tf).
 Home-page: https://github.com/PINTO0309/onnx2tf
 Author: Katsuya Hyodo
@@ -334,7 +334,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  ghcr.io/pinto0309/onnx2tf:1.27.1
+  ghcr.io/pinto0309/onnx2tf:1.27.3
 
   or
 
@@ -342,7 +342,7 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  docker.io/pinto0309/onnx2tf:1.27.1
+  docker.io/pinto0309/onnx2tf:1.27.3
 
   or