Question

代码：

  from keras.applications import InceptionV3
  model = InceptionV3(weights="imagenet")
  shape = (None,image_size,image_size,num_channels)
  x = tf.placeholder(tf.float32, shape=shape)      
  adv_x,grad_x = fgm(x, model, model.predict(x), y=y, targeted=True, eps=0, clip_min=-0.5, clip_max=0.5)
  adv_,grad_ = batch_eval(sess, [x,y], [adv_x,grad_x], [inputs,targets], args={'batch_size': args['batch_size']})

  model.predict(x)

错误：

  File "/u/.../env/lib/python3.5/site-packages/keras/engine/training.py", line 1594, in predict
    batch_size=batch_size, verbose=verbose)
  File "/u/.../env/lib/python3.5/site-packages/keras/engine/training.py", line 1208, in _predict_loop
    batches = _make_batches(samples, batch_size)
  File "/u/.../env/lib/python3.5/site-packages/keras/engine/training.py", line 364, in _make_batches
    num_batches = int(np.ceil(size / float(batch_size)))
TypeError: unsupported operand type(s) for /: 'Dimension' and 'float'

我可以在实际图像上使用model.predict，但最终会在tf.placeholders或tf.variables上出现此错误任何人都可以帮我调试这个错误吗？

Answer 1

Keras＆＃39; Model.predict期望输入数据的numpy数组。除非批量大小为32，否则您可能还需要包含batch_size值。来自文档：

predict(self, x, batch_size=None, verbose=0, steps=None) method of keras.engine.training.Model instance
    Generates output predictions for the input samples.

    Computation is done in batches.

    # Arguments
        x: The input data, as a Numpy array
            (or list of Numpy arrays if the model has multiple outputs).
        batch_size: Integer. If unspecified, it will default to 32.
        verbose: Verbosity mode, 0 or 1.
        steps: Total number of steps (batches of samples)
            before declaring the prediction round finished.
            Ignored with the default value of `None`.

    # Returns
        Numpy array(s) of predictions.

    # Raises
        ValueError: In case of mismatch between the provided
            input data and the model's expectations,

Keras InceptionV3 model.predict

1 个答案: