深度学习新手求助：斗鱼图像分类项目ImageDataGenerator路径错误

Hey there! Let's break down why you're hitting this FileNotFoundError and how to fix it step by step:

1. Fix Path Format Issues #

Your path '\\tmp\Train' has a couple of red flags for Windows systems:

• In Python strings, backslashes act as escape characters. While \T isn't a standard escape sequence, it's far safer to use forward slashes (/) or raw strings to avoid accidental formatting errors.
• Windows doesn't come with a default root-level \tmp directory unless you created it manually. If this folder lives in your project's working directory, use a relative path instead.

Corrected Path Options: #

• Relative path (if tmp/Train is in your current project folder):

  './tmp/Train'
  

• Absolute path (replace with your actual folder location):

  'C:/Users/YourName/Projects/DouyuClassifier/tmp/Train'
  

• Raw string (eliminates escape character risks):

  r'C:\Users\YourName\Projects\DouyuClassifier\tmp\Train'
  

2. Meet Directory Structure Requirements #

flow_from_directory relies on a strict folder structure to categorize images: your Train directory must contain separate subdirectories for each class, with all images for a class stored inside its corresponding subfolder. For binary classification, it should look like this:

tmp/
  Train/
    live_stream/
      img_001.jpg
      img_002.jpg
      ...
    non_live_stream/
      img_001.jpg
      img_002.jpg
      ...

If you place images directly inside Train without these class subfolders, the generator won't find valid class directories—which can also trigger a FileNotFoundError.

3. Verify the Path Actually Exists #

Double-check that the folder path you're using is real:

• Open File Explorer and navigate to the path to confirm it exists.
• Watch for typos (capitalization matters on some Windows setups, and missing slashes will break the path).

Corrected Code Example #

Here's your code with a fixed relative path and proper structure in mind:

from tensorflow.keras.preprocessing.image import ImageDataGenerator

# All images will be rescaled by 1./255
train_datagen = ImageDataGenerator(rescale=1./255)

# Flow training images in batches of 128 using train_datagen generator
train_generator = train_datagen.flow_from_directory(
    './tmp/Train',  # Updated relative path
    target_size=(300, 300),  # Note: Your comment says 150x150 but code uses 300x300—adjust if needed!
    batch_size=128,
    class_mode='binary'
)

Give these fixes a go—they should resolve the FileNotFoundError you're facing!

内容的提问来源于stack exchange，提问作者NANGA PARBAT SUMARNO