为何无法用Image.open打开Hugging Face数据集里的PIL图像？

代码示例 #

import datasets
import PIL
from PIL import Image
from datasets import load_dataset

dataset = load_dataset("ccmusic-database/music_genre", split="test")
output = dataset
im = Image.open(output[0])
im.show()

报错信息 #

Traceback (most recent call last):
File "/Users/abc/Desktop/Project Python Audio/.venv/lib/python3.11/site-packages/PIL/Image.py", line 3247, in open
fp.seek(0)
^^^^^^^
AttributeError: 'dict' object has no attribute 'seek'
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/Users/abc/Desktop/Project Python Audio/loading_dataset.py", line 8, in 
im = Image.open(output[0])
^^^^^^^^^^^^^^^^^^^^^
File "/Users/abc/Desktop/Project Python Audio/.venv/lib/python3.11/site-packages/PIL/Image.py", line 3249, in open
fp = io.BytesIO(fp.read())
^^^^^^^
AttributeError: 'dict' object has no attribute 'read'

打印output[0]的结果 #

{'image': <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=476x349 at 0x101BE6E10>, 'fst_level_label': 1, 'sec_level_label': 6, 'thr_level_label': 6, 'duration': 416.0533106575964}

原因 #

Image.open() 仅能处理文件路径、字节流或类文件对象，用来创建新的PIL图像实例。但你传入的output[0]是一个字典，不符合该函数的输入要求，因此触发报错。实际上这个字典的image字段已经是一个现成的PIL图像对象，无需再次调用Image.open()处理。

解决方法 #

直接提取字典中的image字段，调用其show()方法即可显示图像：

import datasets
from PIL import Image
from datasets import load_dataset

dataset = load_dataset("ccmusic-database/music_genre", split="test")
im = dataset[0]['image']
im.show()

内容的提问来源于stack exchange，提问作者Laulito