Model Folder is empty

[muneebable]

Why the model folder is empty and where is model reside? I can't able to find it

[iamhankai]

I didn't put the pretrained model here. You need to train by yourself.

[muneebable]

Also what is the technique, used for CUB attributes because there are 312 attributes but you processed attributes? What technique you used? I am trying to understand the meaning of this nb_attributes = [10, 16, 16, 16, 5, 16, 7, 16, 12, 16, 16, 15, 4, 16, 16, 16, 16, 6, 6, 15, 5, 5, 5, 16, 16, 16, 16, 5] like what is the meaning of 10 here[position represent bill shape etc]?

[muneebable]

Also, I am getting this error, don't know why?

ValueError                                Traceback (most recent call last)
<ipython-input-8-101167670821> in <module>()
     11 model_raw.summary()
     12 share_fea_map = model_raw.get_layer(shared_layer_name).output
---> 13 share_fea_map = Reshape((final_dim, L), name='reshape_layer')(share_fea_map)
     14 share_fea_map = Permute((2, 1))(share_fea_map)
     15 

C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\topology.py in __call__(self, inputs, **kwargs)
    636             # Inferring the output shape is only relevant for Theano.
    637             if all([s is not None for s in _to_list(input_shape)]):
--> 638                 output_shape = self.compute_output_shape(input_shape)
    639             else:
    640                 if isinstance(input_shape, list):

C:\ProgramData\Anaconda3\lib\site-packages\keras\layers\core.py in compute_output_shape(self, input_shape)
    400             # input shape known? then we can compute the output shape
    401             return (input_shape[0],) + self._fix_unknown_dimension(
--> 402                 input_shape[1:], self.target_shape)
    403 
    404     def call(self, inputs):

C:\ProgramData\Anaconda3\lib\site-packages\keras\layers\core.py in _fix_unknown_dimension(self, input_shape, output_shape)
    388             output_shape[unknown] = original // known
    389         elif original != known:
--> 390             raise ValueError(msg)
    391 
    392         return tuple(output_shape)

ValueError: total size of new array must be unchanged

I am assuming the value of L is wrong because the for RESNET50, activation_49 have shape (?, 2048,7,7) but the while reshaping, you consider the value of L is 14*14.

[iamhankai]

Also what is the technique, used for CUB attributes because there are 312 attributes but you processed attributes? What technique you used? I am trying to understand the meaning of this nb_attributes = [10, 16, 16, 16, 5, 16, 7, 16, 12, 16, 16, 15, 4, 16, 16, 16, 16, 6, 6, 15, 5, 5, 5, 16, 16, 16, 16, 5] like what is the meaning of 10 here[position represent bill shape etc]?

The 312 attributes are clustered according to whether they are the same attribute.

[iamhankai]

Also, I am getting this error, don't know why?

ValueError                                Traceback (most recent call last)
<ipython-input-8-101167670821> in <module>()
     11 model_raw.summary()
     12 share_fea_map = model_raw.get_layer(shared_layer_name).output
---> 13 share_fea_map = Reshape((final_dim, L), name='reshape_layer')(share_fea_map)
     14 share_fea_map = Permute((2, 1))(share_fea_map)
     15 

C:\ProgramData\Anaconda3\lib\site-packages\keras\engine\topology.py in __call__(self, inputs, **kwargs)
    636             # Inferring the output shape is only relevant for Theano.
    637             if all([s is not None for s in _to_list(input_shape)]):
--> 638                 output_shape = self.compute_output_shape(input_shape)
    639             else:
    640                 if isinstance(input_shape, list):

C:\ProgramData\Anaconda3\lib\site-packages\keras\layers\core.py in compute_output_shape(self, input_shape)
    400             # input shape known? then we can compute the output shape
    401             return (input_shape[0],) + self._fix_unknown_dimension(
--> 402                 input_shape[1:], self.target_shape)
    403 
    404     def call(self, inputs):

C:\ProgramData\Anaconda3\lib\site-packages\keras\layers\core.py in _fix_unknown_dimension(self, input_shape, output_shape)
    388             output_shape[unknown] = original // known
    389         elif original != known:
--> 390             raise ValueError(msg)
    391 
    392         return tuple(output_shape)

ValueError: total size of new array must be unchanged

I am assuming the value of L is wrong because the for RESNET50, activation_49 have shape (?, 2048,7,7) but the while reshaping, you consider the value of L is 14*14.

The input image size is 448448, not 224224.

[muneebable]

Thank you for you quick answers.

1) I am getting this error, I dont know what am i doing wrong I exactly run your code same.

Epoch 1/10

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-91-341adc1cba50> in <module>()
     55                 label_batch_list.append(np_utils.to_categorical(y_batch, nb_classes))
     56 
---> 57                 loss = model.train_on_batch(X_batch, label_batch_list)
     58                 # print
     59                 ave_loss = ave_loss*(batches-1)/batches + np.array(loss)/batches

~/anaconda3/envs/3.5pytorch/lib/python3.5/site-packages/keras/engine/training.py in train_on_batch(self, x, y, sample_weight, class_weight)
   1875             x, y,
   1876             sample_weight=sample_weight,
-> 1877             class_weight=class_weight)
   1878         if self.uses_learning_phase and not isinstance(K.learning_phase(), int):
   1879             ins = x + y + sample_weights + [1.]

~/anaconda3/envs/3.5pytorch/lib/python3.5/site-packages/keras/engine/training.py in _standardize_user_data(self, x, y, sample_weight, class_weight, check_array_lengths, batch_size)
   1478                                     output_shapes,
   1479                                     check_batch_axis=False,
-> 1480                                     exception_prefix='target')
   1481         sample_weights = _standardize_sample_weights(sample_weight,
   1482                                                      self._feed_output_names)

~/anaconda3/envs/3.5pytorch/lib/python3.5/site-packages/keras/engine/training.py in _standardize_input_data(data, names, shapes, check_batch_axis, exception_prefix)
    121                             ': expected ' + names[i] + ' to have shape ' +
    122                             str(shape) + ' but got array with shape ' +
--> 123                             str(data_shape))
    124     return data
    125 

ValueError: Error when checking target: expected p0 to have shape (1,) but got array with shape (200,)

[iamhankai]

The categorical_crossentropy loss in Keras may changed its API.

you can try to set from_logits=True in keras.backend.categorical_crossentropy(target, output, from_logits=False, axis=-1).

[muneebable]

The prediction of network is a list of length 30 in which there are 28 attributes and what are the other two? I got the prediction on image, how can i see the attention on image?

[iamhankai]

The region_score_map in https://github.com/iamhankai/attribute-aware-attention/blob/master/cub_demo.py#L112 is the attention values. You can visualize it.