AttributeError: 'SpatialTransformer' object has no attribute 'is_placeholder'

[mdshopon]

I was trying to implement Spatial Transformer Layer in my Code but I am getting this error. Is there any solution for it ?

`
def locnet(): b = np.zeros((2, 3), dtype='float32') b[0, 0] = 1 b[1, 1] = 1 W = np.zeros((64, 6), dtype='float32') weights = [W, b.flatten()] locnet = Sequential()

    locnet.add(Conv2D(16, (7, 7), padding='valid', input_shape=(128, 64, 1)))
    locnet.add(MaxPooling2D(pool_size=(2, 2)))
    locnet.add(Conv2D(32, (5, 5), padding='valid'))
    locnet.add(MaxPooling2D(pool_size=(2, 2)))
    locnet.add(Conv2D(64, (3, 3), padding='valid'))
    locnet.add(MaxPooling2D(pool_size=(2, 2)))

    locnet.add(Flatten())
    locnet.add(Dense(128))
    locnet.add(Activation('elu'))
    locnet.add(Dense(64))
    locnet.add(Activation('elu'))
    locnet.add(Dense(6, weights=weights))

    return locnet

def train(run_name, start_epoch, stop_epoch, img_w):
    # Input Parameters
    img_h = 64
    words_per_epoch = 8000
    val_split = 0.2
    val_words = int(words_per_epoch * (val_split))

    # Network parameters
    conv_filters = 16
    kernel_size = (3, 3)
    pool_size = 2
    time_dense_size = 32
    rnn_size = 512
    minibatch_size = 32

    if K.image_data_format() == 'channels_first':
        input_shape = (1, img_w, img_h)
    else:
        input_shape = (img_w, img_h, 1)

    fdir = os.path.dirname(get_file('wordlists.tgz',
                                    origin='http://www.mythic-ai.com/datasets/wordlists.tgz', untar=True))

    img_gen = TextImageGenerator(monogram_file=os.path.join( 'bangla_wordlist_mono_clean.txt'),
                                 bigram_file=os.path.join(fdir, 'wordlist_bi_clean.txt'),
                                 minibatch_size=minibatch_size,
                                 img_w=img_w,
                                 img_h=img_h,
                                 downsample_factor=(pool_size ** 2),
                                 val_split=words_per_epoch - val_words
                                 )
    act = 'relu'
    input_data = Input(name='the_input', shape=input_shape, dtype='float32')
    input_data=(SpatialTransformer(localization_net=locnet(),
                                 output_size=(128, 64)))(input_data)
    inner = Conv2D(conv_filters, kernel_size, padding='same',
                   activation=act, kernel_initializer='he_normal',
                   name='conv1')(input_data)
    inner = MaxPooling2D(pool_size=(pool_size, pool_size), name='max1')(inner)

    inner = Conv2D(conv_filters, kernel_size, padding='same',
                   activation=act, kernel_initializer='he_normal',
                   name='conv2')(inner)
    inner = MaxPooling2D(pool_size=(pool_size, pool_size), name='max2')(inner)

    conv_to_rnn_dims = (img_w // (pool_size ** 2), (img_h // (pool_size ** 2)) * conv_filters)
    inner = Reshape(target_shape=conv_to_rnn_dims, name='reshape')(inner)

    # cuts down input size going into RNN:
    inner = Dense(time_dense_size, activation=act, name='dense1')(inner)

    # Two layers of bidirectional GRUs
    # GRU seems to work as well, if not better than LSTM:
    gru_1 = GRU(rnn_size, return_sequences=True, kernel_initializer='he_normal', name='gru1')(inner)
    gru_1b = GRU(rnn_size, return_sequences=True, go_backwards=True, kernel_initializer='he_normal', name='gru1_b')(inner)
    gru1_merged = add([gru_1, gru_1b])
    gru_2 = GRU(rnn_size, return_sequences=True, kernel_initializer='he_normal', name='gru2')(gru1_merged)
    gru_2b = GRU(rnn_size, return_sequences=True, go_backwards=True, kernel_initializer='he_normal', name='gru2_b')(gru1_merged)

    # transforms RNN output to character activations:
    inner = Dense(img_gen.get_output_size(), kernel_initializer='he_normal',
                  name='dense2')(concatenate([gru_2, gru_2b]))
    y_pred = Activation('softmax', name='softmax')(inner)
    # Model(inputs=input_data, outputs=y_pred).summary()

    labels = Input(name='the_labels', shape=[img_gen.absolute_max_string_len], dtype='float32')
    input_length = Input(name='input_length', shape=[1], dtype='int64')
    label_length = Input(name='label_length', shape=[1], dtype='int64')
    # Keras doesn't currently support loss funcs with extra parameters
    # so CTC loss is implemented in a lambda layer
    loss_out = Lambda(ctc_lambda_func, output_shape=(1,), name='ctc')([y_pred, labels, input_length, label_length])

    # clipnorm seems to speeds up convergence
    sgd = SGD(lr=0.02, decay=1e-6, momentum=0.9, nesterov=True, clipnorm=5)

    model = Model(inputs=[input_data, labels, input_length, label_length], outputs=loss_out)

    # the loss calc occurs elsewhere, so use a dummy lambda func for the loss
    model.compile(loss={'ctc': lambda y_true, y_pred: y_pred}, optimizer=sgd)`

[hello2all]

Could you post your error output and your tensorflow & keras versions so I can take a look at your problem?

[mdshopon]

Thank you very much for your early reply Keras Version: 2.0.2 Tensorflow Version: 1.3.0

Error:

Traceback (most recent call last): File "bangla_image_ocr.py", line 604, in <module> train(run_name, 0, 40, 128) File "bangla_image_ocr.py", line 580, in train model = Model(inputs=[input_data, labels, input_length, label_length], outputs=loss_out) File "/home/codehead/anaconda2/lib/python2.7/site-packages/keras/legacy/interfaces.py", line 88, in wrapper return func(*args, **kwargs) File "/home/codehead/anaconda2/lib/python2.7/site-packages/keras/engine/topology.py", line 1566, in __init__ if layer.is_placeholder: AttributeError: 'SpatialTransformer' object has no attribute 'is_placeholder'

[hello2all]

Thank you, I'd try to reproduce the error using newer versions of tf and keras to see if the problem is caused by compatibility issues.

[mdshopon]

@hello2all May I know which versions did you use for this ?

[hello2all]

If I recall correctly, tensorflow 1.0 and keras 2.0

[mdshopon]

Thanks ! Please let me know if you have found any solution for this problem .

[hello2all]

@codeheadshopon There is a compatibility issue with the newest version of Keras. However, there is an easy fix to it: in spatia_transformer.py, delete line 37

self.constraints = self.locnet.constraints

After deleting the line, my model was able to successfully compile with no error.

This is tested with: Tensorflow 1.3.0 Keras 2.0.8

[mdshopon]

Didn't work with me. I updated my keras version to 2.0.8 But still getting the same error.

[hello2all]

In the newer versions of Keras, there has been an update on the method to build customized layer. Since I can not re-create the exact error you have been experiencing, I encourage you to look into the Keras documentation and modify the layer initialization accordingly.

I hope this is helpful for you.