Improved conversion stability for tensors with None geometry

README.md

@@ -255,15 +255,15 @@ Video speed is adjusted approximately 50 times slower than actual speed.
   $ docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  ghcr.io/pinto0309/onnx2tf:1.17.8
+  ghcr.io/pinto0309/onnx2tf:1.17.9
 
   or
 
   # Authentication is not required for pulls from Docker Hub.
   $ docker run --rm -it \
   -v `pwd`:/workdir \
   -w /workdir \
-  docker.io/pinto0309/onnx2tf:1.17.8
+  docker.io/pinto0309/onnx2tf:1.17.9
 
   or
 

onnx2tf/__init__.py

@@ -1,3 +1,3 @@
 from onnx2tf.onnx2tf import convert, main
 
-__version__ = '1.17.8'
+__version__ = '1.17.9'

onnx2tf/ops/Concat.py

@@ -215,7 +215,8 @@ def make_node(
 
     # TensorFlow does not support Concat for scalar values, so convert to tensor
     values = [
-        value if len(value.shape) > 0 else tf.reshape(value, [1]) for value in values
+        value if value.shape != tf.TensorShape(None) \
+            and len(value.shape) > 0 else tf.reshape(value, [1]) for value in values
     ]
 
     # Generation of TF OP

onnx2tf/ops/Gather.py

@@ -300,6 +300,9 @@ def make_node(
             and input_tensor.shape[axis] is not None:
             maximum_number_of_elements = input_tensor.shape[axis]
             indices_values = indices_values + maximum_number_of_elements
+        elif tf.keras.backend.is_keras_tensor(indices_values) \
+            and indices_values.shape == tf.TensorShape(None):
+            indices_values = tf.reshape(indices_values, [-1])
 
         tf_layers_dict[graph_node_output.name]['tf_node'] = \
             tf.gather(

onnx2tf/ops/Slice.py

@@ -410,9 +410,13 @@ def make_node(
                 name=graph_node.name,
             )
 
-        check_input_shape = list(input_tensor_shape)
-        check_output_shape = list(tf_layers_dict[graph_node_output.name]['tf_node'].shape)
-        if None not in check_input_shape \
+        check_input_shape = list(input_tensor_shape) \
+            if input_tensor_shape != tf.TensorShape(None) else None
+        check_output_shape = list(tf_layers_dict[graph_node_output.name]['tf_node'].shape) \
+            if tf_layers_dict[graph_node_output.name]['tf_node'].shape != tf.TensorShape(None) else None
+        if check_input_shape is not None \
+            and check_output_shape is not None \
+            and None not in check_input_shape \
             and None not in check_output_shape \
             and check_input_shape == check_output_shape:
             # Disable useless slice