Tensorflow 获取错误“；索引器：列表索引超出范围“；

这是我的密码

model = Sequential()
model.add(Conv2D(32, (3, 3), activation='relu', input_shape=(48, 120, 1)))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(Dropout(0.25))
model.add(MaxPooling2D((2, 2)))
model.add(Conv2D(128, (3, 3), activation='relu'))
model.add(Dropout(0.25))
model.add(MaxPooling2D((2, 2)))
model.add(Conv2D(128, (3, 3), activation='relu'))
model.add(Dropout(0.25))
model.add(MaxPooling2D((2, 2)))
model.add(Flatten())
model.add(Dense(512, activation='relu'))
model.add(Dense(62*4, activation='softmax'))
# print(model.summary())
model.compile(
    loss='categorical_crossentropy',
    optimizer=RMSprop(lr=1e-4),
    metrics=['accuracy']
)
history = model.fit_generator(
    DataGenerator(train_dir, (48, 120, 1)),
    steps_per_epoch=32,
    epochs=30,
    validation_data=DataGenerator(test_dir, (48, 120, 1)),
    validation_steps=32
)

数据发生器

class DataGenerator(Sequence):
    def __init__(self, image_path, shape, to_fit=True, batch_size=32,):
        self.image_path = image_path
        self.shape  = shape
        self.to_fit = to_fit
        self.batch_size = batch_size
        self.all_images = self.get_all_images()
        
    def __len__(self):
        # return the number of batches per epoch
        return len(os.listdir(self.image_path))
        
    def __getitem__(self, index):
        # return a batch of images and (labels) if we are predicting
        if self.to_fit:
            return self._genX_(index), self._genY_(index)
        else:
            return self._genX_(index)
    
    def _genX_(self, index):
        X = self.all_images[index*self.batch_size:(index+1)*self.batch_size]
        X = [cv2.imread(img, 0) for img in X]
        X = [img.reshape(self.shape) for img in X]
        X = [img.astype('float32')/255 for img in X]
        return X
    
    def _genY_(self, index):
        y = self.all_images[index*self.batch_size:(index+1)*self.batch_size]
        return y
    
    def on_epoch_end(self):
        # do something after every epoch
        return
    
    def get_all_images(self):
        return os.listdir(self.image_path)

我得到了什么

Epoch 1/30
Traceback (most recent call last):
  File ".\test_1.py", line 42, in <module>
    validation_steps=32
  File "E:\Environments\Python\Python37\lib\site-packages\keras\legacy\interfaces.py", line 91, in wrapper
    return func(*args, **kwargs)
  File "E:\Environments\Python\Python37\lib\site-packages\keras\engine\training.py", line 1732, in fit_generator
    initial_epoch=initial_epoch)
  File "E:\Environments\Python\Python37\lib\site-packages\keras\engine\training_generator.py", line 220, in fit_generator
    reset_metrics=False)
  File "E:\Environments\Python\Python37\lib\site-packages\keras\engine\training.py", line 1508, in train_on_batch
    class_weight=class_weight)
  File "E:\Environments\Python\Python37\lib\site-packages\keras\engine\training.py", line 579, in _standardize_user_data
    exception_prefix='input')
  File "E:\Environments\Python\Python37\lib\site-packages\keras\engine\training_utils.py", line 87, in standardize_input_data
    if isinstance(data[0], list):
IndexError: list index out of range

1/30纪元
回溯（最近一次呼叫最后一次）：
文件“\test_1.py”，第42行，在
验证步骤=32
文件“E:\Environments\Python\Python37\lib\site packages\keras\legacy\interfaces.py”，第91行，在包装器中
返回函数（*args，**kwargs）
文件“E:\Environments\Python\Python37\lib\site packages\keras\engine\training.py”，第1732行，在fit\u生成器中
初始_历元=初始_历元）
文件“E:\Environments\Python\Python37\lib\site packages\keras\engine\training\u generator.py”，第220行，在fit\u generator中
重置（度量值=False）
文件“E:\Environments\Python\Python37\lib\site packages\keras\engine\training.py”，第1508行，在批处理的列中
等级重量=等级重量）
文件“E:\Environments\Python\Python37\lib\site packages\keras\engine\training.py”，第579行，在用户数据中
异常（前缀为“输入”）
标准化输入数据中的文件“E:\Environments\Python\Python37\lib\site packages\keras\engine\training\u utils.py”，第87行
如果存在（数据[0]，请列出）：
索引器：列表索引超出范围

我只想写我自己的生成器来加载很多图片，但看起来不太好

---

谁能告诉我，我怎么会犯这个错误？如何解决这个问题？

您是否检查

（索引+1）*self.batch\u size

是否低于

len（self.all\u图像）

？已修复，索引*batch\u size太大，找不到图像，我只使用index+batchsize。谢谢