fix bug

x2paddle/op_mapper/paddle_op_mapper.py

@@ -556,12 +556,6 @@ class PaddleOpMapper(object):
                 'SizeTensor' in input_names and
                 len(op.input('SizeTensor')) > 0):
             node_list = list()
-            roi_node = self.make_constant_node(
-                self.get_name(op.type, 'roi'), onnx_pb.TensorProto.FLOAT,
-                [1, 1, 1, 1, 1, 1, 1, 1])
-            roi_name = self.get_name(op.type, 'roi')
-            roi_node = self.make_constant_node(
-                roi_name, onnx_pb.TensorProto.FLOAT, [1, 1, 1, 1, 1, 1, 1, 1])
             empty_name = self.get_name(op.type, 'empty')
             empty_tensor = helper.make_tensor(
                 empty_name,
@@ -584,16 +578,8 @@ class PaddleOpMapper(object):
                 'Slice',
                 inputs=[shape_name0, starts_name, ends_name],
                 outputs=[shape_name1])
-            node_list.extend([
-                roi_node, empty_node, shape_node0, starts_node, ends_node,
-                shape_node1
-            ])
-            #            shape_name2 = self.get_name(op.type, "shape.cast")
-            #            shape_node2 = helper.make_node(
-            #                'Cast',
-            #                inputs=op.input('OutSize'),
-            #                outputs=[shape_name2],
-            #                to=onnx_pb.TensorProto.INT64)
+            node_list.extend(
+                [empty_node, shape_node0, starts_node, ends_node, shape_node1])
             if 'OutSize' in input_names and len(op.input('OutSize')) > 0:
                 cast_shape_name = self.get_name(op.type, "shape.cast")
                 cast_shape_node = helper.make_node(
@@ -603,7 +589,8 @@ class PaddleOpMapper(object):
                     to=onnx_pb.TensorProto.INT64)
                 node_list.append(cast_shape_node)
             else:
-                concat_shape_name = op.output('Out')[0] + "@shape.concat1"
+                concat_shape_name = self.get_name(
+                    op.type, op.output('Out')[0] + "shape.concat")
                 concat_shape_node = helper.make_node(
                     "Concat",
                     inputs=op.input('SizeTensor'),
@@ -616,7 +603,7 @@ class PaddleOpMapper(object):
                     outputs=[cast_shape_name],
                     to=onnx_pb.TensorProto.INT64)
                 node_list.extend([concat_shape_node, cast_shape_node])
-            shape_name3 = op.output('Out')[0] + "@shape.concat3"
+            shape_name3 = self.get_name(op.type, "shape.concat")
             shape_node3 = helper.make_node(
                 'Concat',
                 inputs=[shape_name1, cast_shape_name],
@@ -635,24 +622,24 @@ class PaddleOpMapper(object):
                     vals=[height, width],
                     raw=False))
             node_list.append(node_h_w)
-            outputs_h_w_scales = op.output('Out')[0] + "@out_hw_scales"
-            node_h_w_scales = helper.make_node(
-                'Div',
-                inputs=[shape_name3, name_h_w],
-                outputs=[outputs_h_w_scales])
-            node_list.append(node_h_w_scales)
             shape_name4 = op.output('Out')[0] + "@shape.concat4"
             shape_node4 = helper.make_node(
                 'Concat',
-                inputs=[shape_name1, outputs_h_w_scales],
+                inputs=[shape_name1, name_h_w],
                 outputs=[shape_name4],
                 axis=0)
             node_list.append(shape_node4)
+            outputs_h_w_scales = op.output('Out')[0] + "@out_hw_scales"
+            node_h_w_scales = helper.make_node(
+                'Div',
+                inputs=[shape_name3, shape_name4],
+                outputs=[outputs_h_w_scales])
+            node_list.append(node_h_w_scales)
             result_node = helper.make_node(
                 'Resize',
-                inputs=[op.input('X')[0], shape_name4],
+                inputs=[op.input('X')[0], outputs_h_w_scales],
                 outputs=op.output('Out'),
-                mode='linear', )
+                mode='linear')
             node_list.extend([result_node])
             return node_list
         elif 'Scale' in input_names and len(op.input('Scale')) > 0:
@@ -670,18 +657,12 @@ class PaddleOpMapper(object):
                 scale_node = self.make_constant_node(scale_name,
                                                      onnx_pb.TensorProto.FLOAT,
                                                      [1, 1, scale, scale])
-                roi_name = self.get_name(op.type, 'roi')
-                roi_node = self.make_constant_node(roi_name,
-                                                   onnx_pb.TensorProto.FLOAT,
-                                                   [1, 1, 1, 1, 1, 1, 1, 1])
                 node = helper.make_node(
                     'Resize',
-                    inputs=[op.input('X')[0], roi_name, scale_name],
+                    inputs=[op.input('X')[0], scale_name],
                     outputs=op.output('Out'),
-                    mode='nearest',
-                    coordinate_transformation_mode=coordinate_transformation_mode
-                )
-                return [scale_node, roi_node, node]
+                    mode='nearest')
+                return [scale_node, node]
             else:
                 raise Exception("Unexpected situation happend")
         return node
@@ -713,18 +694,14 @@ class PaddleOpMapper(object):
                 scale_node = self.make_constant_node(scale_name,
                                                      onnx_pb.TensorProto.FLOAT,
                                                      [1, 1, scale, scale])
-                roi_name = self.get_name(op.type, 'roi')
-                roi_node = self.make_constant_node(roi_name,
-                                                   onnx_pb.TensorProto.FLOAT,
-                                                   [1, 1, 1, 1, 1, 1, 1, 1])
                 node = helper.make_node(
                     'Resize',
-                    inputs=[op.input('X')[0], roi_name, scale_name],
+                    inputs=[op.input('X')[0], scale_name],
                     outputs=op.output('Out'),
                     mode='nearest',
                     coordinate_transformation_mode=coordinate_transformation_mode
                 )
-                return [scale_node, roi_node, node]
+                return [scale_node, node]
             else:
                 raise Exception("Unexpected situation happend")
         return node