Keras 的 ImagePreprocessing 类提供了一系列用于图像数据预处理的方法,这些方法可以用于数据增强、标准化等。

主要功能

  • 数据增强:通过旋转、缩放、翻转等方式增加数据多样性。
  • 标准化:将图像数据转换为 [0, 1] 或 [-1, 1] 的范围。

使用方法

from tensorflow.keras.preprocessing.image import ImagePreprocessing

preprocessor = ImagePreprocessing(
    rotation_range=20,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True,
    fill_mode='nearest'
)

# 应用预处理
img = preprocessor.fit_transform(img)

更多关于 Keras 预处理图像的详细使用说明,请参阅官方文档。

扩展阅读

图像预处理示例

的中心图像

ImagePreprocessing 示例

请注意,以上代码示例仅供参考,实际应用时请根据具体需求进行调整。