运行图像特征提取代码时遇ModuleNotFoundError: No module named 'tensorflow.compat'

问题描述 #

编写基于VGG16的图像特征提取代码，预期输出numpy.ndarray类型的特征结果，运行时触发上述模块找不到的错误。

原代码 #

from keras.preprocessing import image
from keras.applications.vgg16 import VGG16 
from keras.applications.vgg16 import preprocess_input 
from keras.models import Model import numpy as np

class FeatureExtractor: def init(self): base_model = VGG16(weights="imagenet") self.model = Model(inputs=base_model.input, outputs=base_model.get_layer("fc1").output)
def extract(self, img):
    img = img.resize((224, 224)).convert("RGB")
    x = image.img_to_array(img)
    x = np.expand_dims(x, axis=0)
    x = preprocess_input(x)
    feature = self.model.predict(x)[0]
    return feature / np.linalg.norm(feature)

错误堆栈 #

Traceback (most recent call last):
  File "C:\Users\Ronte\OneDrive\Documents\Image Search Platfoem\static\offline.py", line 4, in <module>
    from featureextractor import FeatureExtractor
  File "C:\Users\Ronte\OneDrive\Documents\Image Search Platfoem\static\featureextractor.py", line 1, in <module>
    from keras.preprocessing import image
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\__init__.py", line 3, in <module>
    from keras import __internal__
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\__internal__\__init__.py", line 3, in <module>
    from keras.__internal__ import backend
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\__internal__\backend\__init__.py", line 3, in <module>
    from keras.src.backend import _initialize_variables as initialize_variables
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\src\__init__.py", line 21, in <module>
    from keras.src import models
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\src\models\__init__.py", line 18, in <module>
    from keras.src.engine.functional import Functional
  File "C:\Users\Ronte\AppData\Local\Programs\Python\Python310\lib\site-packages\keras\src\engine\functional.py", line 23, in <module>
    import tensorflow.compat.v2 as tf
ModuleNotFoundError: No module named 'tensorflow.compat'

解决方案 #

原因分析 #

错误源于Keras版本（大概率是Keras 3.x）与TensorFlow环境不兼容：Keras 3.x默认依赖TensorFlow作为后端，但你的环境中未安装TensorFlow，或安装的版本过低，导致Keras无法找到tensorflow.compat模块。此外原代码还存在类构造函数命名错误（init应改为__init__）。

修复步骤 #

1. 安装/升级TensorFlow
   打开终端执行命令，安装兼容的TensorFlow版本：

   pip install tensorflow>=2.15.0
   

   若已安装，可升级到最新稳定版：

   pip install --upgrade tensorflow
   

2. 修正代码并适配Keras 3.x
   将导入路径改为tensorflow.keras，同时修复构造函数的命名错误：

   from tensorflow.keras.preprocessing import image
   from tensorflow.keras.applications.vgg16 import VGG16 
   from tensorflow.keras.applications.vgg16 import preprocess_input 
   from tensorflow.keras.models import Model 
   import numpy as np
   
   class FeatureExtractor: 
       def __init__(self):
           base_model = VGG16(weights="imagenet") 
           self.model = Model(inputs=base_model.input, outputs=base_model.get_layer("fc1").output)
       def extract(self, img):
           img = img.resize((224, 224)).convert("RGB")
           x = image.img_to_array(img)
           x = np.expand_dims(x, axis=0)
           x = preprocess_input(x)
           feature = self.model.predict(x)[0]
           return feature / np.linalg.norm(feature)
   

3. 验证环境
   执行以下代码检查TensorFlow和Keras是否正常运行：

   import tensorflow as tf
   from tensorflow.keras.applications.vgg16 import VGG16
   print(tf.__version__)
   print(tf.keras.__version__)
   

   能正常输出版本号则说明环境修复完成。

内容的提问来源于stack exchange，提问作者tg7