Merge branch 'check-non-const-filter' into 'master'

check non-const filter in convolution for onnx model

See merge request !1032

mace/python/tools/converter_tool/onnx_converter.py

@@ -241,14 +241,14 @@ class OnnxNode(object):
         self.node_proto = node
 
     def print_info(self):
-        print "node: ", self.name
-        print "    type: ", self.op_type
-        print "    domain: ", self.domain
-        print "    inputs: ", self.inputs
-        print "    outputs: ", self.outputs
-        print "    attrs:"
+        print("node: ", self.name)
+        print("    type: ", self.op_type)
+        print("    domain: ", self.domain)
+        print("    inputs: ", self.inputs)
+        print("    outputs: ", self.outputs)
+        print("    attrs:")
         for arg in self.attrs:
-            print "        %s: %s" % (arg, self.attrs[arg])
+            print("        %s: %s" % (arg, self.attrs[arg]))
 
 
 class OnnxTensor(object):
@@ -378,11 +378,11 @@ class OnnxConverter(base_converter.ConverterInterface):
         opset_imp = onnx_model.opset_import
 
         polish_available = True
-        print "onnx model IR version: ", ir_version
+        print("onnx model IR version: ", ir_version)
         for imp in opset_imp:
             domain = imp.domain
             version = imp.version
-            print "constains ops domain: ", domain, "version:", version
+            print("constains ops domain: ", domain, "version:", version)
             if 'kaldi2onnx' in domain:
                 polish_available = False
                 self._data_format = DataFormat.DF_NONE
@@ -397,11 +397,11 @@ class OnnxConverter(base_converter.ConverterInterface):
     @staticmethod
     def print_graph_info(graph):
         for value_info in graph.value_info:
-            print "value info:", value_info
+            print("value info:", value_info)
         for value_info in graph.input:
-            print "inputs info:", value_info
+            print("inputs info:", value_info)
         for value_info in graph.output:
-            print "outputs info:", value_info
+            print("outputs info:", value_info)
 
     def extract_shape_info(self, graph):
         def extract_value_info(shape_dict, value_info):
@@ -674,6 +674,8 @@ class OnnxConverter(base_converter.ConverterInterface):
             op.type = MaceOp.DepthwiseConv2d.name
         else:
             op.type = MaceOp.Conv2D.name
+            mace_check(op.input[1] in self._consts,
+                       "Mace does not support non-const filter convolution.")
 
         dilation_arg = op.arg.add()
         dilation_arg.name = MaceKeyword.mace_dilations_str

mace/python/tools/converter_tool/transformer.py

@@ -963,7 +963,9 @@ class Transformer(base_converter.ConverterInterface):
 
         net = self._model
         for op in net.op:
-            if op.type == MaceOp.Conv2D.name:
+            if op.type == MaceOp.Conv2D.name \
+                    and len(op.input) >= 2 \
+                    and op.input[1] in self._consts:
                 producer = self._producer[op.input[0]]
                 input_shape = producer.output_shape[0].dims
                 batch, height, width, channels = self.sort_feature_map_shape(