update onnx coreml convert tool

tools/onnx2tnn/onnx-coreml/coreml2onnx.py

@@ -89,7 +89,10 @@ def main():
     net_output = helper.make_tensor_value_info(net_layers[-1]['top'], TensorProto.FLOAT, [net_output_shape_dict['n'], net_output_shape_dict['k'], net_output_shape_dict['h'], net_output_shape_dict['w']])
     net_outputes = [net_output]
 
-    net_output_name = net_layers[-2]['top']
+    # check the model.espresso.net file to adjust the index #
+    print('check the model.espresso.net file to adjust the index')
+    # net_output_name = net_layers[-2]['top']
+    net_output_name = net_layers[-1]['top']
     if net_output_name.isdigit() != True:
         net_output2_shape_dict = net_layer_shapes[net_output_name]
         net_output2 = helper.make_tensor_value_info(net_output_name, TensorProto.FLOAT, [net_output2_shape_dict['n'], net_output2_shape_dict['k'], net_output2_shape_dict['h'], net_output2_shape_dict['w']])
@@ -106,7 +109,11 @@ def main():
     #创建nodes NodeProto
     onnx_net_nodes = []
     onnx_net_weights = []
-    layer_info = net_layers[1]
+
+    # check the model.espresso.net file to adjust the index #
+    print('check the model.espresso.net file to adjust the index')
+    # layer_info = net_layers[1]
+    layer_info = net_layers[0]
 
     for layer_info in net_layers:
         print(layer_info['type'])
@@ -287,31 +294,60 @@ def main():
                                           )
             onnx_net_nodes.append(layer_node)
         elif layer_info['type'] == 'activation':
+            node_inputs = layer_info['bottom'].split(',')
+            node_outputs = layer_info['top'].split(',')
+
             activation_mode = layer_info['mode']
             if activation_mode == 0:
-                node_type = 'Relu'
+                layer_node = helper.make_node('Relu',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             elif activation_mode == 1:
-                node_type = 'Tanh'
+                layer_node = helper.make_node('Tanh',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             elif activation_mode == 2:
-                node_type = 'LeakyRelu'
+                layer_node = helper.make_node('LeakyRelu',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             elif activation_mode == 3:
-                node_type = 'Sigmoid'
+                layer_node = helper.make_node('Sigmoid',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             elif activation_mode == 4:
-                node_type = 'PRelu'
+                layer_node = helper.make_node('PRelu',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             elif activation_mode == 8:
-                node_type = 'Elu'
+                layer_node = helper.make_node('Elu',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
+            elif activation_mode == 9:
+                layer_node = helper.make_node('ThresholdedRelu',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  alpha = layer_info['alpha']
+                  )
             elif activation_mode == 10:
-                node_type = 'Softplus'
+                layer_node = helper.make_node('Softplus',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
+            elif activation_mode == 12:
+                layer_node = helper.make_node('Softsign',  # node type
+                 node_inputs,  # inputs
+                  node_outputs,  # outputs
+                  )
             else:
                 print('Error: unsupported activation mode: ' + str(activation_mode))
                 assert(0)
 
-            node_inputs = layer_info['bottom'].split(',')
-            node_outputs = layer_info['top'].split(',')
-            layer_node = helper.make_node(node_type,  # node type
-                                          node_inputs,  # inputs
-                                          node_outputs,  # outputs
-                                          )
             onnx_net_nodes.append(layer_node)
         elif layer_info['type'] == 'load_constant':
             # constant_blob
@@ -332,7 +368,7 @@ def main():
 
     #创建model (ModelProto)
     # onnx_model = helper.make_model(graph_def, producer_name='YouTu Tencent')
-    onnx_model = helper.make_model(graph_def, producer_name='YouTu Tencent', opset_imports=[helper.make_operatorsetid("", 9)])
+    onnx_model = helper.make_model(graph_def, producer_name='YouTu Tencent', opset_imports=[helper.make_operatorsetid("", 12)])
 
     # print('The model is:\n{}'.format(onnx_model))
     onnx.checker.check_model(onnx_model)

tools/onnx2tnn/onnx-coreml/onnx2coreml.py

@@ -8,22 +8,22 @@
 import onnx
 from onnx import helper, shape_inference
 # from onnx import version_converter
-from onnx_coreml import convert
 import coremltools
 from coremltools.models import MLModel
 
 def main():
     parser = argparse.ArgumentParser()
-    parser.add_argument('onnx_model_path', help='Input ONNX model path')
+    parser.add_argument('onnx_model_path', help='Input ONNX model full path')
     parser.add_argument('-mlmodelc', help='Save model wiht mlmodelc, 1:default yes, 0: no')
     args = parser.parse_args()
 
     onnx_net_path = args.onnx_model_path
     mlmodelc = args.mlmodelc
     if mlmodelc == None:
-        mlmodelc = 1
+        mlmodelc = '1'
 
-    onnx_net_dir = onnx_net_path[0:onnx_net_path.rfind('/')];
+    onnx_net_dir = onnx_net_path[0:onnx_net_path.rfind('.')];
+    print("dir: "+onnx_net_dir)
 
     cml_net_path = onnx_net_path+'.mlmodel'
     if onnx_net_path.endswith('.onnx'):
@@ -38,13 +38,14 @@ def main():
     # onnx.checker.check_model(onnx_model)
     # onnx_model = shape_inference.infer_shapes(onnx_model)
 
-    cml_model = convert(onnx_model)
+    cml_model = coremltools.converters.onnx.convert(model=onnx_net_path)
     # cml_model = convert(onnx_model, image_input_names=[onnx_inputs[0].name])
     cml_model.save(cml_net_path)
     # new_cml_spec = cml_model.get_spec()
     # coremltools.utils.save_spec(new_cml_spec, cml_net_path)
 
-    if mlmodelc == 1:
+    # print("dir: "+onnx_net_dir)
+    if mlmodelc == '1':
         cmd = '/Applications/Xcode.app/Contents/Developer/usr/bin/coremlc compile '+cml_net_path+' '+onnx_net_dir
         os.system(cmd)
 

tools/onnx2tnn/onnx-coreml/onnx_net.py

@@ -0,0 +1,85 @@
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+
+import os
+import argparse
+import sys
+import time
+
+import struct
+import json
+
+import onnx
+from onnx import helper, shape_inference
+from onnx import AttributeProto, TensorProto, GraphProto
+from onnx import numpy_helper
+
+def main():
+    # parser = argparse.ArgumentParser()
+    # parser.add_argument('coreml_folder', help='Input coreml model folder')
+    # args = parser.parse_args()
+    # coreml_folder = args.coreml_folder
+    #
+    # coreml_net = coreml_folder + '/model.espresso.net'
+    # coreml_shape = coreml_folder + '/model.espresso.shape'
+    # coreml_weights = coreml_folder + '/model.espresso.weights'
+
+    # print(net_layer_data[1])
+    # tensor_shape = [16]
+    # tensor = helper.make_tensor('1', TensorProto.FLOAT, tensor_shape, net_layer_data[1])
+    # print(tensor)
+
+    #构建onnx
+    #创建输入 (ValueInfoProto)
+    net_inputes = []
+    net_input = helper.make_tensor_value_info("input", TensorProto.FLOAT, [1, 13, 4, 4])
+    net_inputes.append(net_input)
+
+    #创建输出 (ValueInfoProto)
+    net_output = helper.make_tensor_value_info("output", TensorProto.FLOAT, [1, 13, 4, 4])
+    net_outputes = [net_output]
+
+    onnx_blob_shapes = [net_input, net_output]
+
+    #创建nodes NodeProto
+    onnx_net_nodes = []
+    onnx_net_weights = []
+
+    node_inputs = ["input"]
+    node_outputs = ["output"]
+    layer_node = helper.make_node("ThresholdedRelu",  # node type
+                                  node_inputs,  # inputs
+                                  node_outputs,  # outputs
+                                  alpha = 1.5
+                                  )
+    onnx_net_nodes.append(layer_node)
+
+    #创建graph GraphProto
+    graph_def = helper.make_graph(
+        onnx_net_nodes,
+        'onnx-model',
+        net_inputes,
+        net_outputes,
+        initializer = onnx_net_weights,
+        value_info = onnx_blob_shapes,
+    )
+
+    #创建model (ModelProto)
+    # onnx_model = helper.make_model(graph_def, producer_name='YouTu Tencent')
+    onnx_model = helper.make_model(graph_def, producer_name='YouTu Tencent', opset_imports=[helper.make_operatorsetid("", 12)])
+
+    # print('The model is:\n{}'.format(onnx_model))
+    onnx.checker.check_model(onnx_model)
+    print('Before shape inference, the shape info of Y is:\n{}'.format(onnx_model.graph.value_info))
+
+    # Apply shape inference on the model
+    inferred_model = shape_inference.infer_shapes(onnx_model)
+    onnx.checker.check_model(inferred_model)
+    print('After shape inference, the shape info of Y is:\n{}'.format(inferred_model.graph.value_info))
+    onnx.save(inferred_model, 'onnx_net.onnx')
+
+
+if __name__ == '__main__':
+    main()

tools/onnx2tnn/onnx-coreml/安装说明-CoreML.md

@@ -16,3 +16,4 @@ export https_proxy=https://web-proxy.oa.com:8080/
 <!-- 编译+运行 -->
 1. 安装coremltools //https://pypi.org/project/coremltools/
 pip3 install -U coremltools
+pip3 install -U onnx-coreml