导入错误：需要 Python 图像库 (PIL) 才能从 jpeg 文件加载数据

Posted 2023-03-12

【中文标题】导入错误：需要 Python 图像库 (PIL) 才能从 jpeg 文件加载数据

【英文标题】：Import Error: The Python Imaging Library (PIL) is required to load data from jpeg files

【发布时间】：2018-08-10 23:59:27

【问题描述】：

我正在尝试使用 scikit-learn 从LFW dataset 获取数据：

from sklearn.datasets import fetch_lfw_people    
faces = fetch_lfw_people(min_faces_per_person=60)

这样做时，我会收到 导入错误 消息：

从 jpeg 文件加载数据需要 Python Imaging Library (PIL)

错误信息表明我需要安装pillow。所以我从我的 Jupyter 笔记本安装了pillow，如下所示：

!conda install --yes --prefix sys.prefix pillow

现在我可以执行import PIL，但在尝试从 LFW 数据集中获取数据时仍然收到相同的错误消息。

我正在使用 Python 3.6。这是完整的跟踪：

ImportError                               Traceback (most recent call last) ~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\datasets\lfw.py in _load_imgs(file_paths, slice_, color, resize)
    143         try:
--> 144             from scipy.misc import imread
    145         except ImportError:

ImportError: cannot import name 'imread'

During handling of the above exception, another exception occurred:

ImportError                               Traceback (most recent call last) ~\AppData\Local\Continuum\anaconda3\lib\site-packages\scipy\misc\pilutil.py in <module>()
     18 try:
---> 19     from PIL import Image, ImageFilter
     20 except ImportError:

~\AppData\Local\Continuum\anaconda3\lib\site-packages\PIL\Image.py in <module>()
     57     # and should be considered private and subject to change.
---> 58     from . import _imaging as core
     59     if PILLOW_VERSION != getattr(core, 'PILLOW_VERSION', None):

ImportError: DLL load failed: The specified module could not be found.

During handling of the above exception, another exception occurred:

ModuleNotFoundError                       Traceback (most recent call last) ~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\datasets\lfw.py in _load_imgs(file_paths, slice_, color, resize)
    145         except ImportError:
--> 146             from scipy.misc.pilutil import imread
    147         from scipy.misc import imresize

~\AppData\Local\Continuum\anaconda3\lib\site-packages\scipy\misc\pilutil.py in <module>()
     20 except ImportError:
---> 21     import Image
     22     import ImageFilter

ModuleNotFoundError: No module named 'Image'

During handling of the above exception, another exception occurred:

ImportError                               Traceback (most recent call last) <ipython-input-236-6d1c601c92a3> in <module>()
----> 1 faces = fetch_lfw_people(min_faces_per_person=60)

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\datasets\lfw.py in fetch_lfw_people(data_home, funneled, resize, min_faces_per_person, color, slice_, download_if_missing)
    333     faces, target, target_names = load_func(
    334         data_folder_path, resize=resize,
--> 335         min_faces_per_person=min_faces_per_person, color=color, slice_=slice_)
    336 
    337     # pack the results as a Bunch instance

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\externals\joblib\memory.py in __call__(self, *args, **kwargs)
    560 
    561     def __call__(self, *args, **kwargs):
--> 562         return self._cached_call(args, kwargs)[0]
    563 
    564     def __reduce__(self):

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\externals\joblib\memory.py in _cached_call(self, args, kwargs)
    508                           'directory %s'
    509                         % (name, argument_hash, output_dir))
--> 510             out, metadata = self.call(*args, **kwargs)
    511             if self.mmap_mode is not None:
    512                 # Memmap the output at the first call to be consistent with

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\externals\joblib\memory.py in call(self, *args, **kwargs)
    742         if self._verbose > 0:
    743             print(format_call(self.func, args, kwargs))
--> 744         output = self.func(*args, **kwargs)
    745         self._persist_output(output, output_dir)
    746         duration = time.time() - start_time

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\datasets\lfw.py in _fetch_lfw_people(data_folder_path, slice_, color, resize, min_faces_per_person)
    234     target = np.searchsorted(target_names, person_names)
    235 
--> 236     faces = _load_imgs(file_paths, slice_, color, resize)
    237 
    238     # shuffle the faces with a deterministic RNG scheme to avoid having

~\AppData\Local\Continuum\anaconda3\lib\site-packages\sklearn\datasets\lfw.py in _load_imgs(file_paths, slice_, color, resize)
    147         from scipy.misc import imresize
    148     except ImportError:
--> 149         raise ImportError("The Python Imaging Library (PIL)"
    150                           " is required to load data from jpeg files")
    151 

ImportError: The Python Imaging Library (PIL) is required to load data from jpeg files

为什么我会收到此导入错误，我该如何解决？

【参考方案1】：

您需要在代码中添加以下行

from PIL import Image

【参考方案2】：

额外细节： Scikit-learn 使用pillow 加载图像。

根据 datasets/base.py 中的 sklearn 代码，pillow 将根据要求导入：

# import PIL only when needed
from ..externals._pilutil import imread

...如果未安装库，则会失败。通过例如安装它conda，执行以下命令：

> conda install --yes --prefix <environment_path_obtained_from_runnig_sys.prefix> pillow

在您重新启动内核/刷新 IDE 以包含新安装后，无需导入库或其任何组件，除非您希望在 sklearn 正在执行的操作之外处理图像。

【参考方案3】：

我遇到了同样的问题。重新启动内核不起作用，但重新启动 Jupyter Lab 可以。我不确定为什么会发生这种情况。