Performs a random rotation of a Numpy image tensor.
tf.keras.preprocessing.image.random_rotation(
x, rg, row_axis=1, col_axis=2, channel_axis=0, fill_mode='nearest',
cval=0.0, interpolation_order=1
)
Arguments |
|---|
x
|
Input tensor. Must be 3D.
|
rg
|
Rotation range, in degrees.
|
row_axis
|
Index of axis for rows in the input tensor.
|
col_axis
|
Index of axis for columns in the input tensor.
|
channel_axis
|
Index of axis for channels in the input tensor.
|
fill_mode
|
Points outside the boundaries of the input
are filled according to the given mode
(one of {'constant', 'nearest', 'reflect', 'wrap'}).
|
cval
|
Value used for points outside the boundaries
of the input if mode='constant'.
|
interpolation_order
|
int, order of spline interpolation.
see ndimage.interpolation.affine_transform
|
Returns |
|---|
|
Rotated Numpy image tensor.
|
TensorFlow 1 version
View source on GitHub