如何在数组中操作数组

Question

我有一个形状为(1433, 1) 的X_train np.array。第一个维度（1433）是训练图像的数量。第二个维度 (1) 是一个 np.array，它本身的形状为 (224, 224, 3)。我可以通过X_train[0][0].shape 确认。我需要将X_train 拟合到模型中：

model.fit([X_train, y_train[:,1:]], y_train[:,0], epochs=50, batch_size=32,  verbose=1)

错误输出不言自明：

    Traceback (most recent call last):
  File "/home/combined/file_01.py", line 97, in <module>
    img_output = Flatten()(x_1)
  File "/usr/local/lib/python3.5/dist-packages/keras/engine/base_layer.py", line 414, in __call__
    self.assert_input_compatibility(inputs)
  File "/usr/local/lib/python3.5/dist-packages/keras/engine/base_layer.py", line 327, in assert_input_compatibility
    str(K.ndim(x)))
ValueError: Input 0 is incompatible with layer flatten_1: expected min_ndim=3, found ndim=2

y_train[:,1:] 似乎可以使用 (1433, 9) 的形状。

model.fit 中的X_train 需要做什么才能成功输入为 (1433, 224, 224, 3)？

Answer 1

你的情况好像是这样的：

import numpy as np
x_train = np.zeros((1433, 1), dtype=object)
for i in range(x_train.shape[0]):
    x_train[i, 0] = np.random.random((224, 224, 3))

x_train.shape        # (1433, 1)
x_train[0, 0].shape  # (224, 224, 3)

其中x_train 是object 数组（如嵌套列表）而不是numeric 数组。

您需要将x_train 更改为纯numeric 数组：

x_train = np.array([x for x in x_train.flatten()], dtype=float)
x_train.shape       # (1433, 224, 224, 3)
x_train[0].shape    # (224, 224, 3)