ValueError when training Tensorflow: setting an array element with a sequence

Question

I was stuck on training process with ValueError. The following are the details:

I made flags as below.

flags = tf.app.flags
FLAGS = flags.FLAGS

flags.DEFINE_string('train', 'train.txt', 'File name of train data')
flags.DEFINE_string('test', 'test.txt', 'File name of train data')
flags.DEFINE_string('train_dir', '/tmp/data', 'Directory to put the training data.')
flags.DEFINE_integer('max_steps', 200, 'Number of steps to run trainer.')
flags.DEFINE_integer('batch_size', 10, 'Batch size'
                     'Must divide evenly into the dataset sizes.')
flags.DEFINE_float('learning_rate', 1e-4, 'Initial learning rate.')

Moreover training process as below:

if 1==1:
        # Tensor for images
    images_placeholder = tf.placeholder("float", shape=(None, IMAGE_PIXELS))
        # Tensor for labels
        labels_placeholder = tf.placeholder("float", shape=(None, NUM_CLASSES))
        # dropout rate
        keep_prob = tf.placeholder("float")

        # call inference() 
        logits = inference(images_placeholder, keep_prob)
        # call loss()
        loss_value = loss(logits, labels_placeholder)
        # call training()
        train_op = training(loss_value, FLAGS.learning_rate)
        # calculate accuract
        acc = accuracy(logits, labels_placeholder)

        # prepare for saving
        saver = tf.train.Saver()
        # make Session
        sess = tf.Session()
        # initialize variables
        sess.run(tf.initialize_all_variables())
        # values on TensorBoard
        summary_op = tf.merge_all_summaries()
        summary_writer = tf.train.SummaryWriter(FLAGS.train_dir, sess.graph_def)

        # Training process
        for step in range(FLAGS.max_steps):
            for i in range(len(train_image)/FLAGS.batch_size):
                # batch_size
                batch = FLAGS.batch_size*i
                # define data in placeholder by feed dict
                sess.run(train_op, feed_dict={
                images_placeholder:train_image[batch:batch+FLAGS.batch_size],
        labels_placeholder: train_label[batch:batch+FLAGS.batch_size],
                keep_prob: 0.5})

When I run this code I faced following error. How to solve this problem?

File "CNN_model.py", line 230, in <module>
    images_placeholder: train_image[batch:batch+FLAGS.batch_size],labels_placeholder: train_label[batch:batch+FLAGS.batch_size],keep_prob: 0.5})
File "/Library/Python/2.7/site-packages/tensorflow/python/client/session.py", line 334, in run
    np_val = np.array(subfeed_val, dtype=subfeed_t.dtype.as_numpy_dtype)
ValueError: setting an array element with a sequence.

I add code around train_image and train_label as following. 　　　

NUM_CLASSES = 5
IMAGE_SIZE = 599
IMAGE_PIXELS = IMAGE_SIZE*1*128

f = open("song_features.json")
 data = json.load(f)
 data = np.array(data)

 flatten_data = []
 flatten_label = []


 for line in range(len(data)):
     for_flat = np.array(data[line])
     flatten_data.append(for_flat.flatten().tolist())

     #label made as 1-of-K
     tmp = np.zeros(NUM_CLASSES)
     tmp[int(random.randint(0,4))] = 1
        flatten_label.append(tmp)

  #1 line training data
  train_image = np.asarray(flatten_data)
  train_label = np.asarray(flatten_label)

I constructing this model.

mrry · Accepted Answer · 2016-02-17 17:38:30Z

3

This exception is raised when TensorFlow converts the values in the feed_dict into dense NumPy ndarrays, and will depend on what's in your train_image and train_label objects.

A common cause of these errors is when feed value is a ragged list: i.e. a list of lists, where the sublists have different sizes. For example:

>>> train_image = [[1., 2., 3.], [4., 5.]]
>>> np.array(train_image, dtype=np.float32)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ValueError: setting an array element with a sequence.

EDIT: Thanks for sharing your code to build train_image and train_label. I suspect the problem is in the creation of train_image, if the elements of flatten_data can have different lengths for each example. Try making the following modification to confirm this: train_image = np.asarray(flatten_data, dtype=np.float32). If you get the same ValueError, you will need to pad or crop the individual rows so that they have IMAGE_PIXELS elements.

edited Feb 17, 2016 at 17:38

answered Feb 17, 2016 at 5:13

mrry

126k27 gold badges403 silver badges401 bronze badges

Sign up to request clarification or add additional context in comments.

4 Comments

koppepanna Over a year ago

Thank you for your support. I added code around train_image and train_label.

mrry Over a year ago

Thanks for sharing the extra code! I updated the answer with some suggestions.

koppepanna Over a year ago

I tried train_image = np.asarray(flatten_data, dtype=np.float32) however, I got the same error. This time I added code around NUM_PIXELS.

mrry Over a year ago

Try adding assert for_flat.flatten().shape == (NUM_PIXELS,) before appending to flatten_data. That should find any images that are the wrong size.

Collectives™ on Stack Overflow

ValueError when training Tensorflow: setting an array element with a sequence

1 Answer 1

4 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

4 Comments

Your Answer

Sign up or log in

Post as a guest

Related