Skip to content

Latest commit

 

History

History
123 lines (95 loc) · 5.43 KB

README.md

File metadata and controls

123 lines (95 loc) · 5.43 KB
Raw

CoreML-Models

Convert AnimeGANv2 to Core ML

AnimeGANv2

  1. Save the graph in pbtxt format.
   tf.train.write_graph(sess.graph_def, './', 'animegan.pbtxt')
  1. Find the name of the output node.
   graph = sess.graph
   print([node.name for node in graph.as_graph_def().node])
  1. Make a frozen graph.
   from tensorflow.python.tools.freeze_graph import freeze_graph
   import tfcoreml

   graph_def_file = 'animegan.pbtxt'
   checkpoint_file = 'checkpoint/generator_Hayao_weight/Hayao-64.ckpt'
   frozen_model_file = './frozen_model.pb'
   output_node_names = 'generator/G_MODEL/out_layer/Tanh'

   freeze_graph(input_graph=graph_def_file,
               input_saver="",
               input_binary=False,
               input_checkpoint=checkpoint_file,
               output_node_names=output_node_names,
               restore_op_name="save/restore_all",
               filename_tensor_name="save/Const:0",
               output_graph=frozen_model_file,
               clear_devices=True,
               initializer_nodes="")
  1. Convert with tfcoreml.
   input_tensor_shapes = {'test:0':[1, 256, 256, 3]} # batch size is 1
   # Output CoreML model path
   coreml_model_file = './animegan.mlmodel'
   output_tensor_names = ['generator/G_MODEL/out_layer/Tanh:0']
   # Call the converter
   coreml_model = tfcoreml.convert(
         tf_model_path='frozen_model.pb',
         mlmodel_path=coreml_model_file,
         input_name_shape_dict=input_tensor_shapes,
         output_feature_names=output_tensor_names,
         image_input_names='test:0',
         red_bias=-1,
         green_bias=-1,
         blue_bias=-1,
         image_scale=2/255,
         minimum_ios_deployment_target='12'
         )

Converted CoreML Models

Image Classifier

Google Drive Link Size Original Project
Efficientnetb0 22.7 MB TensorFlowHub

GAN

Google Drive Link Size Original Project
UGATIT_selfie2anime 1.12GB taki0112/UGATIT
AnimeGANv2_Hayao　 8.7MB TachibanaYoshino/AnimeGANv2
AnimeGANv2_Paprika　 8.7MB TachibanaYoshino/AnimeGANv2
WarpGAN Caricature　 35.5MB seasonSH/WarpGAN
CartoonGAN_Shinkai　 44.6MB mnicnc404/CartoonGan-tensorflow
CartoonGAN_Hayao　 44.6MB mnicnc404/CartoonGan-tensorflow
CartoonGAN_Hosoda　 44.6MB mnicnc404/CartoonGan-tensorflow
CartoonGAN_Paprika　 44.6MB mnicnc404/CartoonGan-tensorflow

スクリーンショット 2020-05-19 11 09 03

How to use in a xcode project.

1,Use CoreGANContainer. You can use models with dragging&dropping into the container project.

2,Or implement Vision request.


import Vision
lazy var coreMLRequest:VNCoreMLRequest = {
   let model = try! VNCoreMLModel(for: modelname().model)
   let request = VNCoreMLRequest(model: model, completionHandler: self.coreMLCompletionHandler)
   return request
   }()

let handler = VNImageRequestHandler(ciImage: ciimage,options: [:])
   DispatchQueue.global(qos: .userInitiated).async {
   try? handler.perform([coreMLRequest])
}

For visualizing multiArray as image, Mr. Hollance’s “CoreML Helpers” are very convenient. CoreML Helpers

Converting from MultiArray to Image with CoreML Helpers.

func coreMLCompletionHandler（request：VNRequest？、error：Error？）{
   let = coreMLRequest.results？.first as！VNCoreMLFeatureValueObservation
   let multiArray = result.featureValue.multiArrayValue
   let cgimage = multiArray？.cgImage（min：-1、max：1、channel：nil）

Apps made by Core ML models. AnimateU