Question

我正在尝试使用feed_dict喂神经网络，但这会产生此错误“ 不可散列的类型：'numpy.ndarray'”

提要字典的输入是图像，图像是形状（宽度，高度，通道）和方向的图像列表，方向是2d数组

def batch_gen(data_dir, image_paths, steering_angles, batch_size, 
    is_training):
    index = np.random.permutation(image_paths.shape[0])
    for center, left, right in image_paths[index]:
       center = center
       left = left
       right = right
       break
    steering_angle = steering_angles[index]
    # argumentation
    if is_training and np.random.rand() < 0.6:
        image, steering_angle = augument(data_dir, center, left, right, 
         steering_angle)
    else:
        image = load_image(data_dir, center) 
            # add the image and steering angle to the batch
    images = preprocess(image)
    steers = steering_angle
    return images,steers

#

with tf.Session() as sess:

# Run the initializer
    sess.run(tf.global_variables_initializer())
    for step in range(1, num_steps+1):
            # Run optimization op (backprop)
        images, steer = batch_gen(data_dir, X_train, y_train, 5, True)
        print(images.shape)
        sess.run(optimizer, feed_dict={images, steer})

那么不可散列的含义是什么，我该如何解决这个问题

Answer 1

feed_dict是一个字典（键值对）。

e.g. feed_dict={  x: images, y: steer }

x＆y是必须为可哈希类型的键。在您的情况下，您直接将图像作为字典键传递，这会导致产生无法哈希的类型错误。

x＆y（对于您的网络，名称可能有所不同）通常是网络中的tf.placeholder。

例如

import tensorflow as tf

x = tf.placeholder("float", None)
y = tf.placeholder("float", None)
z = x * y

with tf.Session() as session:
    result = session.run(z, feed_dict={x: [1, 2, 3], y: [2,4,6]})
    print(result)

无法将图片送入feed_dict

1 个答案: