图像生成器中的错误：要求检索元素 0，但序列的长度为 0

Posted 2023-02-16

【中文标题】图像生成器中的错误：要求检索元素 0，但序列的长度为 0

【英文标题】：Error in Image generator : Asked to retrieve element 0, but the Sequence has length 0

【发布时间】：2021-02-04 15:28:44

【问题描述】：

我正在尝试计算二进制类彩色图像分类问题中的真阳性、真阴性、假阳性、假阴性比率。

我有二进制类、人脸和背景彩色图像，我必须使用 MLP 对它们进行分类。

我的问题是：我得到错误：

ValueError: 要求检索元素 0，但序列的长度为 0

编辑：完整追溯

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py in _method_wrapper(self, *args, **kwargs)
    128       raise ValueError(' is not supported in multi-worker mode.'.format(
    129           method.__name__))
--> 130     return method(self, *args, **kwargs)
    131 
    132   return tf_decorator.make_decorator(

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/training.py in predict(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)
   1577           use_multiprocessing=use_multiprocessing,
   1578           model=self,
-> 1579           steps_per_execution=self._steps_per_execution)
   1580 
   1581       # Container that configures and calls `tf.keras.Callback`s.

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py in __init__(self, x, y, sample_weight, batch_size, steps_per_epoch, initial_epoch, epochs, shuffle, class_weight, max_queue_size, workers, use_multiprocessing, model, steps_per_execution)
   1115         use_multiprocessing=use_multiprocessing,
   1116         distribution_strategy=ds_context.get_strategy(),
-> 1117         model=model)
   1118 
   1119     strategy = ds_context.get_strategy()

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py in __init__(self, x, y, sample_weights, shuffle, workers, use_multiprocessing, max_queue_size, model, **kwargs)
    914         max_queue_size=max_queue_size,
    915         model=model,
--> 916         **kwargs)
    917 
    918   @staticmethod

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py in __init__(self, x, y, sample_weights, workers, use_multiprocessing, max_queue_size, model, **kwargs)
    784     # Since we have to know the dtype of the python generator when we build the
    785     # dataset, we have to look at a batch to infer the structure.
--> 786     peek, x = self._peek_and_restore(x)
    787     peek = self._standardize_batch(peek)
    788     peek = _process_tensorlike(peek)

/usr/local/lib/python3.6/dist-packages/tensorflow/python/keras/engine/data_adapter.py in _peek_and_restore(x)
    918   @staticmethod
    919   def _peek_and_restore(x):
--> 920     return x[0], x
    921 
    922   def _handle_multiprocessing(self, x, workers, use_multiprocessing,

/usr/local/lib/python3.6/dist-packages/keras_preprocessing/image/iterator.py in __getitem__(self, idx)
     55                              'but the Sequence '
     56                              'has length length'.format(idx=idx,
---> 57                                                           length=len(self)))
     58         if self.seed is not None:
     59             np.random.seed(self.seed + self.total_batches_seen)

ValueError: Asked to retrieve element 0, but the Sequence has length 0

在尝试分别预测 2 个类中的每个文件夹时

（而不是包含 2 个文件夹的根文件夹，每个类一个用于训练）

。

我产生错误的代码是：

test_face_dir = "/content/test/TESTSET/face"
test_background_dir = "/content/test/TESTSET/background"
# Face DG
test_datagen_face = ImageDataGenerator(rescale=1./255)

test_generator_face = test_datagen_face.flow_from_directory(
    test_face_dir,
    target_size=img_window[:2],
    batch_size=batch_size,
    class_mode='binary',
    color_mode='rgb'
    )
# Background DG
test_datagen_background = ImageDataGenerator(rescale=1./255)

test_generator_background = test_datagen_background.flow_from_directory(
    test_background_dir,
    target_size=img_window[:2],
    batch_size=batch_size,
    class_mode='binary',
    color_mode='rgb'
    )
#-----------------------------------------
prediction_face = simpleMLP.predict(test_generator_face)
prediction_background = simpleMLP.predict(test_generator_background)
#-----------------------------------------
# th = 0.5 #threshold
# Face
prediction_face[prediction_face>=th]=1 
prediction_face[prediction_face<th]=0
pred_face = np.squeeze(prediction_face)
print('pred shape: ', pred_face.shape,int(np.sum(pred_face)))
# Background
prediction_background[prediction_background>=th]=1 
prediction_background[prediction_background<th]=0
pred_background = np.squeeze(prediction_background)
print('pred shape: ', pred_background.shape,int(np.sum(pred_background)))

【问题讨论】：

请注意，我们无法运行您的代码，因为我们没有输入数据。帮助我们帮助您并包含完整的回溯，格式为代码。

@MisterMiyagi Full Traceback 添加。

回溯似乎只显示库函数。它不包含您的任何代码。这真的是整个追溯吗？

@MisterMiyagi 是的，完整的追溯。我选择了另一种方法来完成这项工作，但是这种将数据生成器用于 1 个文件夹而不是不同的 2 个文件夹的方法不起作用。

【参考方案1】：

生成错误是因为我使用的是 class_mode='binary'，而我指定的是一个图像文件夹而不是两个文件夹 faces 和 backgrounds ，当我选择包含它们的父文件夹时，这个问题就消失了。

无论如何计算我得到的 True Positive、True Negative、False Positive、False Negative 比率基本事实如下：

# Get Y_GroundTruth
y_label = (np.expand_dims(test_generator_eval.classes, axis=1).ravel()).astype(int)