将batchdataset与keras vgg16 precrocess_input连接

Question

我正在使用 tf.keras.preprocessing.image_dataset_from_directory 获取 batchdataset ，其中数据集有10个类。

我正在尝试将此 batchdataset 与keras vgg16 （ docs ）网络。从文档中：

注意：每个KERAS应用程序都期望特定的输入预处理。对于VGG16，请致电 tf.keras.applications.vgg16.preprocess_input 在将它们传递给模型之前。

但是，我正在努力获得此 preprocess_input 与 batchdataset一起工作。 您能帮我弄清楚如何连接这两个点吗？

请参阅以下代码：

train_ds = tf.keras.preprocessing.image_dataset_from_directory(train_data_dir, image_size=(224, 224))
train_ds = tf.keras.applications.vgg16.preprocess_input(train_ds)

这将抛出 typeerror：'batchdataset'对象无法订阅：

Traceback (most recent call last):
  ...
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/vgg16.py", line 232, in preprocess_input
    return imagenet_utils.preprocess_input(
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/imagenet_utils.py", line 117, in preprocess_input
    return _preprocess_symbolic_input(
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/imagenet_utils.py", line 278, in _preprocess_symbolic_input
    x = x[..., ::-1]
TypeError: 'BatchDataset' object is not subscriptable

来自 typeError：'datasetV1adapter'对象不可订阅（来自 batchdataset试图将python字典作为表格）提示是使用：

train_ds = tf.keras.applications.vgg16.preprocess_input(
    list(train_ds.as_numpy_iterator())
)

但是，这也失败了：这

Traceback (most recent call last):
  ...
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/vgg16.py", line 232, in preprocess_input
    return imagenet_utils.preprocess_input(
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/imagenet_utils.py", line 117, in preprocess_input
    return _preprocess_symbolic_input(
  File "/path/to/venv/lib/python3.10/site-packages/keras/applications/imagenet_utils.py", line 278, in _preprocess_symbolic_input
    x = x[..., ::-1]
TypeError: list indices must be integers or slices, not tuple

一切都失败了：这都是使用 python = 3.10 == 3.10 == 3.10 .3 带有 tensorflow == 2.8.0 。

我该如何工作？先感谢您。

Answer 1

好吧，我弄清楚了。我需要传递 tf.tensor ，而不是 tf.data.dataset 。可以通过迭代数据集来获得张量。

这可以通过几种方式完成：

train_ds = tf.keras.preprocessing.image_dataset_from_directory(...)

# Option 1
batch_images = next(iter(train_ds))[0]
preprocessed_images = tf.keras.applications.vgg16.preprocess_input(batch_images)

# Option 2:
for batch_images, batch_labels in train_ds:
    preprocessed_images = tf.keras.applications.vgg16.preprocess_input(batch_images)

如果将选项2转换为生成器，则可以直接传递到下游 model.fit 中。干杯!