tensorflow2转tflite提示OP不支持的解决方案

爱被打了一巴掌 2022-10-11 13:46 1601阅读 0赞

## 环境 ##

*  Python 3.7.10
 *  TensorFlow 2.5.0

转换代码

import tensorflow as tf
    
    model_dir = './ckpt-120'
    # Convert the model
    converter = tf.lite.TFLiteConverter.from_saved_model(model_dir) # path to the SavedModel directory
    tflite_model = converter.convert()
    
    # Save the model.
    with open('model.tflite', 'wb') as f:
      f.write(tflite_model)

但在这里报错：

<unknown>:0: error: failed while converting: 'main':
    Some ops are not supported by the native TFLite runtime, you can enable TF kernels fallback using TF Select. See instructions: https://www.tensorflow.org/lite/guide/ops_select
    TF Select ops: Conv2D, Mul
    Details:
    	tf.Conv2D {data_format = "NHWC", dilations = [1, 1, 1, 1], explicit_paddings = [0, 0, 1, 1, 1, 1, 0, 0], padding = "EXPLICIT", strides = [1, 2, 2, 1], use_cudnn_on_gpu = true}
    	tf.Mul {device = ""}

现在只需加一行代码就可解决

import tensorflow as tf
    
    model_dir = './ckpt-120'
    # Convert the model
    converter = tf.lite.TFLiteConverter.from_saved_model(model_dir) # path to the SavedModel directory
    converter.target_spec.supported_ops = [
            tf.lite.OpsSet.TFLITE_BUILTINS, tf.lite.OpsSet.SELECT_TF_OPS
    ]
    tflite_model = converter.convert()
    
    # Save the model.
    with open('model.tflite', 'wb') as f:
      f.write(tflite_model)

如果还不行，加上以下几行

converter =  tf.lite.TFLiteConverter.from_saved_model(model_dir)
    converter.experimental_new_converter = False
    converter.optimizations = [tf.lite.Optimize.OPTIMIZE_FOR_SIZE]
    converter.target_spec.supported_ops = [tf.lite.OpsSet.TFLITE_BUILTINS, tf.lite.OpsSet.SELECT_TF_OPS]
    converter.allow_custom_ops=True
    tflite_model = converter.convert()