Convert a list of images and labels to np array to train tensorflow

    image_batch = tf.train.batch([resized_image], batch_size=100)

images, label_batch = tf.train.batch(
    [image, label],
    batch_size=batch_size,
    num_threads=num_preprocess_threads,
    capacity=min_queue_examples + 3 * batch_size)

tf.one_hot(labels, depth=2)

array([[ 0.,  1.],
       [ 0.,  1.],
       [ 0.,  1.],
       [ 1.,  0.],
       [ 1.,  0.]], dtype=float32)

<tf.Tensor 'one_hot:0' shape=(5, 2) dtype=float32>

labels = tf.placeholder(tf.int32, [None])  # 'None' means it has one dimension that is determined by your batch size
# ... define your network ...
loss_op = tf.nn.softmax_cross_entropy_with_logits(logits=logits,
                                                  labels=tf.one_hot(labels))
loss_op = tf.reduce_mean(loss_op)

for _ in range(num_iter):
    d = # generate data batch
    t = # generate label batch, e.g. [1, 1] for the first two images
    _, batch_loss = sess.run([train_op, loss_op],
                             feed_dict={data: d, labels: t})

# define universe of possible input values
alphabet = 'abcdefghijklmnopqrstuvwxyz'

# define a mapping of chars to integers
char_to_int = dict((c, i) for i, c in enumerate(alphabet))
int_to_char = dict((i, c) for i, c in enumerate(alphabet))


def one_hot_encode(data_array):
    integer_encoded = [char_to_int[char] for char in data_array]

    # one hot encode
    onehot_encoded = list()
    for value in integer_encoded:
        letter = [0 for _ in range(len(alphabet))]
        letter[value] = 1
        onehot_encoded.append(letter)

    return onehot_encoded

  a = np.array([[1, 2], [3, 4]])
  b = np.array([[5, 6]])
  np.concatenate((a, b), axis=0)

array([[1, 2],
       [3, 4],
       [5, 6]])