tf.keras.backend.pool3d

View source on GitHub

3D Pooling.

tf.keras.backend.pool3d(
    x, pool_size, strides=(1, 1, 1), padding='valid', data_format=None,
    pool_mode='max'
)

Arguments:

• x: Tensor or variable.
• pool_size: tuple of 3 integers.
• strides: tuple of 3 integers.
• padding: string, "same" or "valid".
• data_format: string, "channels_last" or "channels_first".
• pool_mode: string, "max" or "avg".

Returns:

A tensor, result of 3D pooling.

Raises:

• ValueError: if data_format is neither "channels_last" or "channels_first".
• ValueError: if pool_mode is neither "max" or "avg".