torchvision.transforms¶
Transforms are common image transformations. They can be chained together using Compose
.
Additionally, there is the torchvision.transforms.functional
module.
Functional transforms give finegrained control over the transformations.
This is useful if you have to build a more complex transformation pipeline
(e.g. in the case of segmentation tasks).
All transformations accept PIL Image, Tensor Image or batch of Tensor Images as input. Tensor Image is a tensor with
(C, H, W)
shape, where C
is a number of channels, H
and W
are image height and width. Batch of
Tensor Images is a tensor of (B, C, H, W)
shape, where B
is a number of images in the batch. Deterministic or
random transformations applied on the batch of Tensor Images identically transform all the images of the batch.
Warning
Since v0.8.0 all random transformations are using torch default random generator to sample random parameters. It is a backward compatibility breaking change and user should set the random state as following:
# Previous versions
# import random
# random.seed(12)
# Now
import torch
torch.manual_seed(17)
Please, keep in mind that the same seed for torch random generator and Python random generator will not produce the same results.
Scriptable transforms¶
In order to script the transformations, please use torch.nn.Sequential
instead of Compose
.
transforms = torch.nn.Sequential(
transforms.CenterCrop(10),
transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)),
)
scripted_transforms = torch.jit.script(transforms)
Make sure to use only scriptable transformations, i.e. that work with torch.Tensor
and does not require
lambda functions or PIL.Image
.
For any custom transformations to be used with torch.jit.script
, they should be derived from torch.nn.Module
.
Compositions of transforms¶

class
torchvision.transforms.
Compose
(transforms)[source]¶ Composes several transforms together. This transform does not support torchscript. Please, see the note below.
Parameters: transforms (list of Transform
objects) – list of transforms to compose.Example
>>> transforms.Compose([ >>> transforms.CenterCrop(10), >>> transforms.ToTensor(), >>> ])
Note
In order to script the transformations, please use
torch.nn.Sequential
as below.>>> transforms = torch.nn.Sequential( >>> transforms.CenterCrop(10), >>> transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)), >>> ) >>> scripted_transforms = torch.jit.script(transforms)
Make sure to use only scriptable transformations, i.e. that work with
torch.Tensor
, does not require lambda functions orPIL.Image
.
Transforms on PIL Image and torch.*Tensor¶

class
torchvision.transforms.
CenterCrop
(size)[source]¶ Crops the given image at the center. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters: size (sequence or int) – Desired output size of the crop. If size is an int instead of sequence like (h, w), a square crop (size, size) is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).

class
torchvision.transforms.
ColorJitter
(brightness=0, contrast=0, saturation=0, hue=0)[source]¶ Randomly change the brightness, contrast and saturation of an image.
Parameters:  brightness (float or tuple of python:float (min, max)) – How much to jitter brightness. brightness_factor is chosen uniformly from [max(0, 1  brightness), 1 + brightness] or the given [min, max]. Should be non negative numbers.
 contrast (float or tuple of python:float (min, max)) – How much to jitter contrast. contrast_factor is chosen uniformly from [max(0, 1  contrast), 1 + contrast] or the given [min, max]. Should be non negative numbers.
 saturation (float or tuple of python:float (min, max)) – How much to jitter saturation. saturation_factor is chosen uniformly from [max(0, 1  saturation), 1 + saturation] or the given [min, max]. Should be non negative numbers.
 hue (float or tuple of python:float (min, max)) – How much to jitter hue. hue_factor is chosen uniformly from [hue, hue] or the given [min, max]. Should have 0<= hue <= 0.5 or 0.5 <= min <= max <= 0.5.

class
torchvision.transforms.
FiveCrop
(size)[source]¶ Crop the given image into four corners and the central crop. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Note
This transform returns a tuple of images and there may be a mismatch in the number of inputs and targets your Dataset returns. See below for an example of how to deal with this.
Parameters: size (sequence or int) – Desired output size of the crop. If size is an int
instead of sequence like (h, w), a square crop of size (size, size) is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).Example
>>> transform = Compose([ >>> FiveCrop(size), # this is a list of PIL Images >>> Lambda(lambda crops: torch.stack([ToTensor()(crop) for crop in crops])) # returns a 4D tensor >>> ]) >>> #In your test loop you can do the following: >>> input, target = batch # input is a 5d tensor, target is 2d >>> bs, ncrops, c, h, w = input.size() >>> result = model(input.view(1, c, h, w)) # fuse batch size and ncrops >>> result_avg = result.view(bs, ncrops, 1).mean(1) # avg over crops

class
torchvision.transforms.
Grayscale
(num_output_channels=1)[source]¶ Convert image to grayscale. The image can be a PIL Image or a Tensor, in which case it is expected to have […, 3, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters: num_output_channels (int) – (1 or 3) number of channels desired for output image Returns:  Grayscale version of the input.
 If
num_output_channels == 1
: returned image is single channel  If
num_output_channels == 3
: returned image is 3 channel with r == g == b
 If
Return type: PIL Image

class
torchvision.transforms.
Pad
(padding, fill=0, padding_mode='constant')[source]¶ Pad the given image on all sides with the given “pad” value. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  padding (int or tuple or list) – Padding on each border. If a single int is provided this
is used to pad all borders. If tuple of length 2 is provided this is the padding
on left/right and top/bottom respectively. If a tuple of length 4 is provided
this is the padding for the left, top, right and bottom borders respectively.
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:
[padding, ]
.  fill (int or tuple) – Pixel fill value for constant fill. Default is 0. If a tuple of length 3, it is used to fill R, G, B channels respectively. This value is only used when the padding_mode is constant
 padding_mode (str) –
Type of padding. Should be: constant, edge, reflect or symmetric. Default is constant. Mode symmetric is not yet supported for Tensor inputs.
 constant: pads with a constant value, this value is specified with fill
 edge: pads with the last value at the edge of the image
 reflect: pads with reflection of image without repeating the last value on the edgeFor example, padding [1, 2, 3, 4] with 2 elements on both sides in reflect mode will result in [3, 2, 1, 2, 3, 4, 3, 2]
 symmetric: pads with reflection of image repeating the last value on the edgeFor example, padding [1, 2, 3, 4] with 2 elements on both sides in symmetric mode will result in [2, 1, 1, 2, 3, 4, 4, 3]
 padding (int or tuple or list) – Padding on each border. If a single int is provided this
is used to pad all borders. If tuple of length 2 is provided this is the padding
on left/right and top/bottom respectively. If a tuple of length 4 is provided
this is the padding for the left, top, right and bottom borders respectively.
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:

class
torchvision.transforms.
RandomAffine
(degrees, translate=None, scale=None, shear=None, resample=0, fillcolor=0)[source]¶ Random affine transformation of the image keeping center invariant. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  degrees (sequence or float or int) – Range of degrees to select from. If degrees is a number instead of sequence like (min, max), the range of degrees will be (degrees, +degrees). Set to 0 to deactivate rotations.
 translate (tuple, optional) – tuple of maximum absolute fraction for horizontal and vertical translations. For example translate=(a, b), then horizontal shift is randomly sampled in the range img_width * a < dx < img_width * a and vertical shift is randomly sampled in the range img_height * b < dy < img_height * b. Will not translate by default.
 scale (tuple, optional) – scaling factor interval, e.g (a, b), then scale is randomly sampled from the range a <= scale <= b. Will keep original scale by default.
 shear (sequence or float or int, optional) – Range of degrees to select from. If shear is a number, a shear parallel to the x axis in the range (shear, +shear) will be applied. Else if shear is a tuple or list of 2 values a shear parallel to the x axis in the range (shear[0], shear[1]) will be applied. Else if shear is a tuple or list of 4 values, a xaxis shear in (shear[0], shear[1]) and yaxis shear in (shear[2], shear[3]) will be applied. Will not apply shear by default.
 resample (int, optional) – An optional resampling filter. See filters for more information.
If omitted, or if the image has mode “1” or “P”, it is set to
PIL.Image.NEAREST
. If input is Tensor, onlyPIL.Image.NEAREST
andPIL.Image.BILINEAR
are supported.  fillcolor (tuple or int) – Optional fill color (Tuple for RGB Image and int for grayscale) for the area outside the transform in the output image (Pillow>=5.0.0). This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.

forward
(img)[source]¶  img (PIL Image or Tensor): Image to be transformed.
Returns: Affine transformed image. Return type: PIL Image or Tensor

static
get_params
(degrees: List[float], translate: Union[List[float], NoneType], scale_ranges: Union[List[float], NoneType], shears: Union[List[float], NoneType], img_size: List[int]) → Tuple[float, Tuple[int, int], float, Tuple[float, float]][source]¶ Get parameters for affine transformation
Returns: params to be passed to the affine transformation

class
torchvision.transforms.
RandomApply
(transforms, p=0.5)[source]¶ Apply randomly a list of transformations with a given probability.
Note
In order to script the transformation, please use
torch.nn.ModuleList
as input instead of list/tuple of transforms as shown below:>>> transforms = transforms.RandomApply(torch.nn.ModuleList([ >>> transforms.ColorJitter(), >>> ]), p=0.3) >>> scripted_transforms = torch.jit.script(transforms)
Make sure to use only scriptable transformations, i.e. that work with
torch.Tensor
, does not require lambda functions orPIL.Image
.Parameters:

class
torchvision.transforms.
RandomCrop
(size, padding=None, pad_if_needed=False, fill=0, padding_mode='constant')[source]¶ Crop the given image at a random location. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  size (sequence or int) – Desired output size of the crop. If size is an int instead of sequence like (h, w), a square crop (size, size) is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).
 padding (int or sequence, optional) – Optional padding on each border
of the image. Default is None. If a single int is provided this
is used to pad all borders. If tuple of length 2 is provided this is the padding
on left/right and top/bottom respectively. If a tuple of length 4 is provided
this is the padding for the left, top, right and bottom borders respectively.
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:
[padding, ]
.  pad_if_needed (boolean) – It will pad the image if smaller than the desired size to avoid raising an exception. Since cropping is done after padding, the padding seems to be done at a random offset.
 fill (int or tuple) – Pixel fill value for constant fill. Default is 0. If a tuple of length 3, it is used to fill R, G, B channels respectively. This value is only used when the padding_mode is constant
 padding_mode (str) –
Type of padding. Should be: constant, edge, reflect or symmetric. Default is constant. Mode symmetric is not yet supported for Tensor inputs.
 constant: pads with a constant value, this value is specified with fill
 edge: pads with the last value on the edge of the image
 reflect: pads with reflection of image (without repeating the last value on the edge)padding [1, 2, 3, 4] with 2 elements on both sides in reflect mode will result in [3, 2, 1, 2, 3, 4, 3, 2]
 symmetric: pads with reflection of image (repeating the last value on the edge)padding [1, 2, 3, 4] with 2 elements on both sides in symmetric mode will result in [2, 1, 1, 2, 3, 4, 4, 3]

forward
(img)[source]¶ Parameters: img (PIL Image or Tensor) – Image to be cropped. Returns: Cropped image. Return type: PIL Image or Tensor

static
get_params
(img: torch.Tensor, output_size: Tuple[int, int]) → Tuple[int, int, int, int][source]¶ Get parameters for
crop
for a random crop.Parameters:  img (PIL Image or Tensor) – Image to be cropped.
 output_size (tuple) – Expected output size of the crop.
Returns: params (i, j, h, w) to be passed to
crop
for random crop.Return type:

class
torchvision.transforms.
RandomGrayscale
(p=0.1)[source]¶ Randomly convert image to grayscale with a probability of p (default 0.1). The image can be a PIL Image or a Tensor, in which case it is expected to have […, 3, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters: p (float) – probability that image should be converted to grayscale. Returns: Grayscale version of the input image with probability p and unchanged with probability (1p).  If input image is 1 channel: grayscale version is 1 channel  If input image is 3 channel: grayscale version is 3 channel with r == g == b Return type: PIL Image or Tensor

class
torchvision.transforms.
RandomHorizontalFlip
(p=0.5)[source]¶ Horizontally flip the given image randomly with a given probability. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters: p (float) – probability of the image being flipped. Default value is 0.5

class
torchvision.transforms.
RandomPerspective
(distortion_scale=0.5, p=0.5, interpolation=2, fill=0)[source]¶ Performs a random perspective transformation of the given image with a given probability. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  distortion_scale (float) – argument to control the degree of distortion and ranges from 0 to 1. Default is 0.5.
 p (float) – probability of the image being transformed. Default is 0.5.
 interpolation (int) – Interpolation type. If input is Tensor, only
PIL.Image.NEAREST
andPIL.Image.BILINEAR
are supported. Default,PIL.Image.BILINEAR
for PIL images and Tensors.  fill (ntuple or int or float) – Pixel fill value for area outside the rotated
image. If int or float, the value is used for all bands respectively. Default is 0.
This option is only available for
pillow>=5.0.0
. This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.

forward
(img)[source]¶ Parameters: img (PIL Image or Tensor) – Image to be Perspectively transformed. Returns: Randomly transformed image. Return type: PIL Image or Tensor

static
get_params
(width: int, height: int, distortion_scale: float) → Tuple[List[List[int]], List[List[int]]][source]¶ Get parameters for
perspective
for a random perspective transform.Parameters: Returns: List containing [topleft, topright, bottomright, bottomleft] of the original image, List containing [topleft, topright, bottomright, bottomleft] of the transformed image.

class
torchvision.transforms.
RandomResizedCrop
(size, scale=(0.08, 1.0), ratio=(0.75, 1.3333333333333333), interpolation=2)[source]¶ Crop the given image to random size and aspect ratio. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
A crop of random size (default: of 0.08 to 1.0) of the original size and a random aspect ratio (default: of 3/4 to 4/3) of the original aspect ratio is made. This crop is finally resized to given size. This is popularly used to train the Inception networks.
Parameters:  size (int or sequence) – expected output size of each edge. If size is an
int instead of sequence like (h, w), a square output size
(size, size)
is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).  scale (tuple of python:float) – range of size of the origin size cropped
 ratio (tuple of python:float) – range of aspect ratio of the origin aspect ratio cropped.
 interpolation (int) – Desired interpolation enum defined by filters.
Default is
PIL.Image.BILINEAR
. If input is Tensor, onlyPIL.Image.NEAREST
,PIL.Image.BILINEAR
andPIL.Image.BICUBIC
are supported.

forward
(img)[source]¶ Parameters: img (PIL Image or Tensor) – Image to be cropped and resized. Returns: Randomly cropped and resized image. Return type: PIL Image or Tensor
 size (int or sequence) – expected output size of each edge. If size is an
int instead of sequence like (h, w), a square output size

class
torchvision.transforms.
RandomRotation
(degrees, resample=False, expand=False, center=None, fill=None)[source]¶ Rotate the image by angle. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  degrees (sequence or float or int) – Range of degrees to select from. If degrees is a number instead of sequence like (min, max), the range of degrees will be (degrees, +degrees).
 resample (int, optional) – An optional resampling filter. See filters for more information.
If omitted, or if the image has mode “1” or “P”, it is set to PIL.Image.NEAREST.
If input is Tensor, only
PIL.Image.NEAREST
andPIL.Image.BILINEAR
are supported.  expand (bool, optional) – Optional expansion flag. If true, expands the output to make it large enough to hold the entire rotated image. If false or omitted, make the output image the same size as the input image. Note that the expand flag assumes rotation around the center and no translation.
 center (list or tuple, optional) – Optional center of rotation, (x, y). Origin is the upper left corner. Default is the center of the image.
 fill (ntuple or int or float) – Pixel fill value for area outside the rotated image. If int or float, the value is used for all bands respectively. Defaults to 0 for all bands. This option is only available for Pillow>=5.2.0. This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.

class
torchvision.transforms.
RandomSizedCrop
(*args, **kwargs)[source]¶ Note: This transform is deprecated in favor of RandomResizedCrop.

class
torchvision.transforms.
RandomVerticalFlip
(p=0.5)[source]¶ Vertically flip the given image randomly with a given probability. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters: p (float) – probability of the image being flipped. Default value is 0.5

class
torchvision.transforms.
Resize
(size, interpolation=2)[source]¶ Resize the input image to the given size. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  size (sequence or int) – Desired output size. If size is a sequence like
(h, w), output size will be matched to this. If size is an int,
smaller edge of the image will be matched to this number.
i.e, if height > width, then image will be rescaled to
(size * height / width, size).
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:
[size, ]
.  interpolation (int, optional) – Desired interpolation enum defined by filters.
Default is
PIL.Image.BILINEAR
. If input is Tensor, onlyPIL.Image.NEAREST
,PIL.Image.BILINEAR
andPIL.Image.BICUBIC
are supported.
 size (sequence or int) – Desired output size. If size is a sequence like
(h, w), output size will be matched to this. If size is an int,
smaller edge of the image will be matched to this number.
i.e, if height > width, then image will be rescaled to
(size * height / width, size).
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:

class
torchvision.transforms.
Scale
(*args, **kwargs)[source]¶ Note: This transform is deprecated in favor of Resize.

class
torchvision.transforms.
TenCrop
(size, vertical_flip=False)[source]¶ Crop the given image into four corners and the central crop plus the flipped version of these (horizontal flipping is used by default). The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Note
This transform returns a tuple of images and there may be a mismatch in the number of inputs and targets your Dataset returns. See below for an example of how to deal with this.
Parameters: Example
>>> transform = Compose([ >>> TenCrop(size), # this is a list of PIL Images >>> Lambda(lambda crops: torch.stack([ToTensor()(crop) for crop in crops])) # returns a 4D tensor >>> ]) >>> #In your test loop you can do the following: >>> input, target = batch # input is a 5d tensor, target is 2d >>> bs, ncrops, c, h, w = input.size() >>> result = model(input.view(1, c, h, w)) # fuse batch size and ncrops >>> result_avg = result.view(bs, ncrops, 1).mean(1) # avg over crops

class
torchvision.transforms.
GaussianBlur
(kernel_size, sigma=(0.1, 2.0))[source]¶ Blurs image with randomly chosen Gaussian blur. The image can be a PIL Image or a Tensor, in which case it is expected to have […, C, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  kernel_size (int or sequence) – Size of the Gaussian kernel.
 sigma (float or tuple of python:float (min, max)) – Standard deviation to be used for creating kernel to perform blurring. If float, sigma is fixed. If it is tuple of float (min, max), sigma is chosen uniformly at random to lie in the given range.
Returns: Gaussian blurred version of the input image.
Return type: PIL Image or Tensor

forward
(img: torch.Tensor) → torch.Tensor[source]¶ Parameters: img (PIL Image or Tensor) – image to be blurred. Returns: Gaussian blurred image Return type: PIL Image or Tensor
Transforms on PIL Image only¶
Transforms on torch.*Tensor only¶

class
torchvision.transforms.
LinearTransformation
(transformation_matrix, mean_vector)[source]¶ Transform a tensor image with a square transformation matrix and a mean_vector computed offline. Given transformation_matrix and mean_vector, will flatten the torch.*Tensor and subtract mean_vector from it which is then followed by computing the dot product with the transformation matrix and then reshaping the tensor to its original shape.
 Applications:
 whitening transformation: Suppose X is a column vector zerocentered data. Then compute the data covariance matrix [D x D] with torch.mm(X.t(), X), perform SVD on this matrix and pass it as transformation_matrix.
Parameters:  transformation_matrix (Tensor) – tensor [D x D], D = C x H x W
 mean_vector (Tensor) – tensor [D], D = C x H x W

class
torchvision.transforms.
Normalize
(mean, std, inplace=False)[source]¶ Normalize a tensor image with mean and standard deviation. Given mean:
(mean[1],...,mean[n])
and std:(std[1],..,std[n])
forn
channels, this transform will normalize each channel of the inputtorch.*Tensor
i.e.,output[channel] = (input[channel]  mean[channel]) / std[channel]
Note
This transform acts out of place, i.e., it does not mutate the input tensor.
Parameters:  mean (sequence) – Sequence of means for each channel.
 std (sequence) – Sequence of standard deviations for each channel.
 inplace (bool,optional) – Bool to make this operation inplace.

class
torchvision.transforms.
RandomErasing
(p=0.5, scale=(0.02, 0.33), ratio=(0.3, 3.3), value=0, inplace=False)[source]¶ Randomly selects a rectangle region in an image and erases its pixels. ‘Random Erasing Data Augmentation’ by Zhong et al. See https://arxiv.org/abs/1708.04896
Parameters:  p – probability that the random erasing operation will be performed.
 scale – range of proportion of erased area against input image.
 ratio – range of aspect ratio of erased area.
 value – erasing value. Default is 0. If a single int, it is used to erase all pixels. If a tuple of length 3, it is used to erase R, G, B channels respectively. If a str of ‘random’, erasing each pixel with random values.
 inplace – boolean to make this transform inplace. Default set to False.
Returns: Erased Image.
Example
>>> transform = transforms.Compose([ >>> transforms.RandomHorizontalFlip(), >>> transforms.ToTensor(), >>> transforms.Normalize((0.485, 0.456, 0.406), (0.229, 0.224, 0.225)), >>> transforms.RandomErasing(), >>> ])

forward
(img)[source]¶ Parameters: img (Tensor) – Tensor image to be erased. Returns: Erased Tensor image. Return type: img (Tensor)

static
get_params
(img: torch.Tensor, scale: Tuple[float, float], ratio: Tuple[float, float], value: Union[List[float], NoneType] = None) → Tuple[int, int, int, int, torch.Tensor][source]¶ Get parameters for
erase
for a random erasing.Parameters:  img (Tensor) – Tensor image to be erased.
 scale (tuple or list) – range of proportion of erased area against input image.
 ratio (tuple or list) – range of aspect ratio of erased area.
 value (list, optional) – erasing value. If None, it is interpreted as “random”
(erasing each pixel with random values). If
len(value)
is 1, it is interpreted as a number, i.e.value[0]
.
Returns: params (i, j, h, w, v) to be passed to
erase
for random erasing.Return type:

class
torchvision.transforms.
ConvertImageDtype
(dtype: torch.dtype) → None[source]¶ Convert a tensor image to the given
dtype
and scale the values accordinglyParameters: dtype (torch.dpython:type) – Desired data type of the output Note
When converting from a smaller to a larger integer
dtype
the maximum values are not mapped exactly. If converted back and forth, this mismatch has no effect.Raises: RuntimeError
– When trying to casttorch.float32
totorch.int32
ortorch.int64
as well as for trying to casttorch.float64
totorch.int64
. These conversions might lead to overflow errors since the floating pointdtype
cannot store consecutive integers over the whole range of the integerdtype
.
Conversion Transforms¶

class
torchvision.transforms.
ToPILImage
(mode=None)[source]¶ Convert a tensor or an ndarray to PIL Image. This transform does not support torchscript.
Converts a torch.*Tensor of shape C x H x W or a numpy ndarray of shape H x W x C to a PIL Image while preserving the value range.
Parameters: mode (PIL.Image mode) – color space and pixel depth of input data (optional). If mode
isNone
(default) there are some assumptions made about the input data:  If the input has 4 channels, themode
is assumed to beRGBA
.  If the input has 3 channels, themode
is assumed to beRGB
.  If the input has 2 channels, themode
is assumed to beLA
.  If the input has 1 channel, themode
is determined by the data type (i.eint
,float
,short
).

class
torchvision.transforms.
ToTensor
[source]¶ Convert a
PIL Image
ornumpy.ndarray
to tensor. This transform does not support torchscript.Converts a PIL Image or numpy.ndarray (H x W x C) in the range [0, 255] to a torch.FloatTensor of shape (C x H x W) in the range [0.0, 1.0] if the PIL Image belongs to one of the modes (L, LA, P, I, F, RGB, YCbCr, RGBA, CMYK, 1) or if the numpy.ndarray has dtype = np.uint8
In the other cases, tensors are returned without scaling.
Note
Because the input image is scaled to [0.0, 1.0], this transformation should not be used when transforming target image masks. See the references for implementing the transforms for image masks.
Generic Transforms¶
Functional Transforms¶
Functional transforms give you finegrained control of the transformation pipeline. As opposed to the transformations above, functional transforms don’t contain a random number generator for their parameters. That means you have to specify/generate all parameters, but you can reuse the functional transform.
Example: you can apply a functional transform with the same parameters to multiple images like this:
import torchvision.transforms.functional as TF
import random
def my_segmentation_transforms(image, segmentation):
if random.random() > 0.5:
angle = random.randint(30, 30)
image = TF.rotate(image, angle)
segmentation = TF.rotate(segmentation, angle)
# more transforms ...
return image, segmentation
Example: you can use a functional transform to build transform classes with custom behavior:
import torchvision.transforms.functional as TF
import random
class MyRotationTransform:
"""Rotate by one of the given angles."""
def __init__(self, angles):
self.angles = angles
def __call__(self, x):
angle = random.choice(self.angles)
return TF.rotate(x, angle)
rotation_transform = MyRotationTransform(angles=[30, 15, 0, 15, 30])

torchvision.transforms.functional.
adjust_brightness
(img: torch.Tensor, brightness_factor: float) → torch.Tensor[source]¶ Adjust brightness of an Image.
Parameters:  img (PIL Image or Tensor) – Image to be adjusted.
 brightness_factor (float) – How much to adjust the brightness. Can be any non negative number. 0 gives a black image, 1 gives the original image while 2 increases the brightness by a factor of 2.
Returns: Brightness adjusted image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
adjust_contrast
(img: torch.Tensor, contrast_factor: float) → torch.Tensor[source]¶ Adjust contrast of an Image.
Parameters:  img (PIL Image or Tensor) – Image to be adjusted.
 contrast_factor (float) – How much to adjust the contrast. Can be any non negative number. 0 gives a solid gray image, 1 gives the original image while 2 increases the contrast by a factor of 2.
Returns: Contrast adjusted image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
adjust_gamma
(img: torch.Tensor, gamma: float, gain: float = 1) → torch.Tensor[source]¶ Perform gamma correction on an image.
Also known as Power Law Transform. Intensities in RGB mode are adjusted based on the following equation:
\[I_{\text{out}} = 255 \times \text{gain} \times \left(\frac{I_{\text{in}}}{255}\right)^{\gamma}\]See Gamma Correction for more details.
Parameters: Returns: Gamma correction adjusted image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
adjust_hue
(img: torch.Tensor, hue_factor: float) → torch.Tensor[source]¶ Adjust hue of an image.
The image hue is adjusted by converting the image to HSV and cyclically shifting the intensities in the hue channel (H). The image is then converted back to original image mode.
hue_factor is the amount of shift in H channel and must be in the interval [0.5, 0.5].
See Hue for more details.
Parameters:  img (PIL Image or Tensor) – Image to be adjusted.
 hue_factor (float) – How much to shift the hue channel. Should be in [0.5, 0.5]. 0.5 and 0.5 give complete reversal of hue channel in HSV space in positive and negative direction respectively. 0 means no shift. Therefore, both 0.5 and 0.5 will give an image with complementary colors while 0 gives the original image.
Returns: Hue adjusted image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
adjust_saturation
(img: torch.Tensor, saturation_factor: float) → torch.Tensor[source]¶ Adjust color saturation of an image.
Parameters:  img (PIL Image or Tensor) – Image to be adjusted.
 saturation_factor (float) – How much to adjust the saturation. 0 will give a black and white image, 1 will give the original image while 2 will enhance the saturation by a factor of 2.
Returns: Saturation adjusted image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
affine
(img: torch.Tensor, angle: float, translate: List[int], scale: float, shear: List[float], resample: int = 0, fillcolor: Union[int, NoneType] = None) → torch.Tensor[source]¶ Apply affine transformation on the image keeping image center invariant. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  img (PIL Image or Tensor) – image to transform.
 angle (float or int) – rotation angle in degrees between 180 and 180, clockwise direction.
 translate (list or tuple of python:integers) – horizontal and vertical translations (postrotation translation)
 scale (float) – overall scale
 shear (float or tuple or list) – shear angle value in degrees between 180 to 180, clockwise direction. If a tuple of list is specified, the first value corresponds to a shear parallel to the x axis, while the second value corresponds to a shear parallel to the y axis.
 resample (
PIL.Image.NEAREST
orPIL.Image.BILINEAR
orPIL.Image.BICUBIC
, optional) – An optional resampling filter. See filters for more information. If omitted, or if the image is PIL Image and has mode “1” or “P”, it is set toPIL.Image.NEAREST
. If input is Tensor, onlyPIL.Image.NEAREST
andPIL.Image.BILINEAR
are supported.  fillcolor (int) – Optional fill color for the area outside the transform in the output image (Pillow>=5.0.0). This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.
Returns: Transformed image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
center_crop
(img: torch.Tensor, output_size: List[int]) → torch.Tensor[source]¶ Crops the given image at the center. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  img (PIL Image or Tensor) – Image to be cropped.
 output_size (sequence or int) – (height, width) of the crop box. If int or sequence with single int it is used for both directions.
Returns: Cropped image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
convert_image_dtype
(image: torch.Tensor, dtype: torch.dtype = torch.float32) → torch.Tensor[source]¶ Convert a tensor image to the given
dtype
and scale the values accordinglyParameters:  image (torch.Tensor) – Image to be converted
 dtype (torch.dpython:type) – Desired data type of the output
Returns: Converted image
Return type: Tensor
Note
When converting from a smaller to a larger integer
dtype
the maximum values are not mapped exactly. If converted back and forth, this mismatch has no effect.Raises: RuntimeError
– When trying to casttorch.float32
totorch.int32
ortorch.int64
as well as for trying to casttorch.float64
totorch.int64
. These conversions might lead to overflow errors since the floating pointdtype
cannot store consecutive integers over the whole range of the integerdtype
.

torchvision.transforms.functional.
crop
(img: torch.Tensor, top: int, left: int, height: int, width: int) → torch.Tensor[source]¶ Crop the given image at specified location and output size. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  img (PIL Image or Tensor) – Image to be cropped. (0,0) denotes the top left corner of the image.
 top (int) – Vertical component of the top left corner of the crop box.
 left (int) – Horizontal component of the top left corner of the crop box.
 height (int) – Height of the crop box.
 width (int) – Width of the crop box.
Returns: Cropped image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
erase
(img: torch.Tensor, i: int, j: int, h: int, w: int, v: torch.Tensor, inplace: bool = False) → torch.Tensor[source]¶ Erase the input Tensor Image with given value.
Parameters:  img (Tensor Image) – Tensor image of size (C, H, W) to be erased
 i (int) – i in (i,j) i.e coordinates of the upper left corner.
 j (int) – j in (i,j) i.e coordinates of the upper left corner.
 h (int) – Height of the erased region.
 w (int) – Width of the erased region.
 v – Erasing value.
 inplace (bool, optional) – For inplace operations. By default is set False.
Returns: Erased image.
Return type: Tensor Image

torchvision.transforms.functional.
five_crop
(img: torch.Tensor, size: List[int]) → Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor][source]¶ Crop the given image into four corners and the central crop. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Note
This transform returns a tuple of images and there may be a mismatch in the number of inputs and targets your
Dataset
returns.Parameters:  img (PIL Image or Tensor) – Image to be cropped.
 size (sequence or int) – Desired output size of the crop. If size is an int instead of sequence like (h, w), a square crop (size, size) is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).
Returns:  tuple (tl, tr, bl, br, center)
Corresponding top left, top right, bottom left, bottom right and center crop.
Return type:

torchvision.transforms.functional.
gaussian_blur
(img: torch.Tensor, kernel_size: List[int], sigma: Union[List[float], NoneType] = None) → torch.Tensor[source]¶ Performs Gaussian blurring on the img by given kernel. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  img (PIL Image or Tensor) – Image to be blurred
 kernel_size (sequence of python:ints or int) – Gaussian kernel size. Can be a sequence of integers
like
(kx, ky)
or a single integer for square kernels. In torchscript mode kernel_size as single int is not supported, use a tuple or list of length 1:[ksize, ]
.  sigma (sequence of python:floats or float, optional) – Gaussian kernel standard deviation. Can be a
sequence of floats like
(sigma_x, sigma_y)
or a single float to define the same sigma in both X/Y directions. If None, then it is computed usingkernel_size
assigma = 0.3 * ((kernel_size  1) * 0.5  1) + 0.8
. Default, None. In torchscript mode sigma as single float is not supported, use a tuple or list of length 1:[sigma, ]
.
Returns: Gaussian Blurred version of the image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
hflip
(img: torch.Tensor) → torch.Tensor[source]¶ Horizontally flip the given PIL Image or Tensor.
Parameters: img (PIL Image or Tensor) – Image to be flipped. If img is a Tensor, it is expected to be in […, H, W] format, where … means it can have an arbitrary number of trailing dimensions. Returns: Horizontally flipped image. Return type: PIL Image or Tensor

torchvision.transforms.functional.
normalize
(tensor: torch.Tensor, mean: List[float], std: List[float], inplace: bool = False) → torch.Tensor[source]¶ Normalize a tensor image with mean and standard deviation.
Note
This transform acts out of place by default, i.e., it does not mutates the input tensor.
See
Normalize
for more details.Parameters:  tensor (Tensor) – Tensor image of size (C, H, W) or (B, C, H, W) to be normalized.
 mean (sequence) – Sequence of means for each channel.
 std (sequence) – Sequence of standard deviations for each channel.
 inplace (bool,optional) – Bool to make this operation inplace.
Returns: Normalized Tensor image.
Return type: Tensor

torchvision.transforms.functional.
pad
(img: torch.Tensor, padding: List[int], fill: int = 0, padding_mode: str = 'constant') → torch.Tensor[source]¶ Pad the given image on all sides with the given “pad” value. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  img (PIL Image or Tensor) – Image to be padded.
 padding (int or tuple or list) – Padding on each border. If a single int is provided this
is used to pad all borders. If tuple of length 2 is provided this is the padding
on left/right and top/bottom respectively. If a tuple of length 4 is provided
this is the padding for the left, top, right and bottom borders respectively.
In torchscript mode padding as single int is not supported, use a tuple or
list of length 1:
[padding, ]
.  fill (int or str or tuple) – Pixel fill value for constant fill. Default is 0. If a tuple of length 3, it is used to fill R, G, B channels respectively. This value is only used when the padding_mode is constant. Only int value is supported for Tensors.
 padding_mode –
Type of padding. Should be: constant, edge, reflect or symmetric. Default is constant. Mode symmetric is not yet supported for Tensor inputs.
 constant: pads with a constant value, this value is specified with fill
 edge: pads with the last value on the edge of the image
 reflect: pads with reflection of image (without repeating the last value on the edge)padding [1, 2, 3, 4] with 2 elements on both sides in reflect mode will result in [3, 2, 1, 2, 3, 4, 3, 2]
 symmetric: pads with reflection of image (repeating the last value on the edge)padding [1, 2, 3, 4] with 2 elements on both sides in symmetric mode will result in [2, 1, 1, 2, 3, 4, 4, 3]
Returns: Padded image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
perspective
(img: torch.Tensor, startpoints: List[List[int]], endpoints: List[List[int]], interpolation: int = 2, fill: Union[int, NoneType] = None) → torch.Tensor[source]¶ Perform perspective transform of the given image. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  img (PIL Image or Tensor) – Image to be transformed.
 startpoints (list of list of python:ints) – List containing four lists of two integers corresponding to four corners
[topleft, topright, bottomright, bottomleft]
of the original image.  endpoints (list of list of python:ints) – List containing four lists of two integers corresponding to four corners
[topleft, topright, bottomright, bottomleft]
of the transformed image.  interpolation (int) – Interpolation type. If input is Tensor, only
PIL.Image.NEAREST
andPIL.Image.BILINEAR
are supported. Default,PIL.Image.BILINEAR
for PIL images and Tensors.  fill (ntuple or int or float) – Pixel fill value for area outside the rotated
image. If int or float, the value is used for all bands respectively.
This option is only available for
pillow>=5.0.0
. This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.
Returns: transformed Image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
pil_to_tensor
(pic)[source]¶ Convert a
PIL Image
to a tensor of the same type.See
PILToTensor
for more details.Parameters: pic (PIL Image) – Image to be converted to tensor. Returns: Converted image. Return type: Tensor

torchvision.transforms.functional.
resize
(img: torch.Tensor, size: List[int], interpolation: int = 2) → torch.Tensor[source]¶ Resize the input image to the given size. The image can be a PIL Image or a torch Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Parameters:  img (PIL Image or Tensor) – Image to be resized.
 size (sequence or int) – Desired output size. If size is a sequence like
(h, w), the output size will be matched to this. If size is an int,
the smaller edge of the image will be matched to this number maintaining
the aspect ratio. i.e, if height > width, then image will be rescaled to
\(\left(\text{size} \times \frac{\text{height}}{\text{width}}, \text{size}\right)\).
In torchscript mode size as single int is not supported, use a tuple or
list of length 1:
[size, ]
.  interpolation (int, optional) – Desired interpolation enum defined by filters.
Default is
PIL.Image.BILINEAR
. If input is Tensor, onlyPIL.Image.NEAREST
,PIL.Image.BILINEAR
andPIL.Image.BICUBIC
are supported.
Returns: Resized image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
resized_crop
(img: torch.Tensor, top: int, left: int, height: int, width: int, size: List[int], interpolation: int = 2) → torch.Tensor[source]¶ Crop the given image and resize it to desired size. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Notably used in
RandomResizedCrop
.Parameters:  img (PIL Image or Tensor) – Image to be cropped. (0,0) denotes the top left corner of the image.
 top (int) – Vertical component of the top left corner of the crop box.
 left (int) – Horizontal component of the top left corner of the crop box.
 height (int) – Height of the crop box.
 width (int) – Width of the crop box.
 size (sequence or int) – Desired output size. Same semantics as
resize
.  interpolation (int, optional) – Desired interpolation enum defined by filters.
Default is
PIL.Image.BILINEAR
. If input is Tensor, onlyPIL.Image.NEAREST
,PIL.Image.BILINEAR
andPIL.Image.BICUBIC
are supported.
Returns: Cropped image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
rgb_to_grayscale
(img: torch.Tensor, num_output_channels: int = 1) → torch.Tensor[source]¶ Convert RGB image to grayscale version of image. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Note
Please, note that this method supports only RGB images as input. For inputs in other color spaces, please, consider using meth:~torchvision.transforms.functional.to_grayscale with PIL Image.
Parameters:  img (PIL Image or Tensor) – RGB Image to be converted to grayscale.
 num_output_channels (int) – number of channels of the output image. Value can be 1 or 3. Default, 1.
Returns:  Grayscale version of the image.
if num_output_channels = 1 : returned image is single channel
if num_output_channels = 3 : returned image is 3 channel with r = g = b
Return type: PIL Image or Tensor

torchvision.transforms.functional.
rotate
(img: torch.Tensor, angle: float, resample: int = 0, expand: bool = False, center: Union[List[int], NoneType] = None, fill: Union[int, NoneType] = None) → torch.Tensor[source]¶ Rotate the image by angle. The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions.
Parameters:  img (PIL Image or Tensor) – image to be rotated.
 angle (float or int) – rotation angle value in degrees, counterclockwise.
 resample (
PIL.Image.NEAREST
orPIL.Image.BILINEAR
orPIL.Image.BICUBIC
, optional) – An optional resampling filter. See filters for more information. If omitted, or if the image has mode “1” or “P”, it is set toPIL.Image.NEAREST
.  expand (bool, optional) – Optional expansion flag. If true, expands the output image to make it large enough to hold the entire rotated image. If false or omitted, make the output image the same size as the input image. Note that the expand flag assumes rotation around the center and no translation.
 center (list or tuple, optional) – Optional center of rotation. Origin is the upper left corner. Default is the center of the image.
 fill (ntuple or int or float) – Pixel fill value for area outside the rotated
image. If int or float, the value is used for all bands respectively.
Defaults to 0 for all bands. This option is only available for
pillow>=5.2.0
. This option is not supported for Tensor input. Fill value for the area outside the transform in the output image is always 0.
Returns: Rotated image.
Return type: PIL Image or Tensor

torchvision.transforms.functional.
ten_crop
(img: torch.Tensor, size: List[int], vertical_flip: bool = False) → List[torch.Tensor][source]¶ Generate ten cropped images from the given image. Crop the given image into four corners and the central crop plus the flipped version of these (horizontal flipping is used by default). The image can be a PIL Image or a Tensor, in which case it is expected to have […, H, W] shape, where … means an arbitrary number of leading dimensions
Note
This transform returns a tuple of images and there may be a mismatch in the number of inputs and targets your
Dataset
returns.Parameters:  img (PIL Image or Tensor) – Image to be cropped.
 size (sequence or int) – Desired output size of the crop. If size is an int instead of sequence like (h, w), a square crop (size, size) is made. If provided a tuple or list of length 1, it will be interpreted as (size[0], size[0]).
 vertical_flip (bool) – Use vertical flipping instead of horizontal
Returns:  tuple (tl, tr, bl, br, center, tl_flip, tr_flip, bl_flip, br_flip, center_flip)
Corresponding top left, top right, bottom left, bottom right and center crop and same for the flipped image.
Return type:

torchvision.transforms.functional.
to_grayscale
(img, num_output_channels=1)[source]¶ Convert PIL image of any mode (RGB, HSV, LAB, etc) to grayscale version of image.
Parameters:  img (PIL Image) – PIL Image to be converted to grayscale.
 num_output_channels (int) – number of channels of the output image. Value can be 1 or 3. Default, 1.
Returns:  Grayscale version of the image.
if num_output_channels = 1 : returned image is single channel
if num_output_channels = 3 : returned image is 3 channel with r = g = b
Return type: PIL Image

torchvision.transforms.functional.
to_pil_image
(pic, mode=None)[source]¶ Convert a tensor or an ndarray to PIL Image.
See
ToPILImage
for more details.Parameters:  pic (Tensor or numpy.ndarray) – Image to be converted to PIL Image.
 mode (PIL.Image mode) – color space and pixel depth of input data (optional).
Returns: Image converted to PIL Image. Return type: PIL Image

torchvision.transforms.functional.
to_tensor
(pic)[source]¶ Convert a
PIL Image
ornumpy.ndarray
to tensor.See
ToTensor
for more details.Parameters: pic (PIL Image or numpy.ndarray) – Image to be converted to tensor. Returns: Converted image. Return type: Tensor

torchvision.transforms.functional.
vflip
(img: torch.Tensor) → torch.Tensor[source]¶ Vertically flip the given PIL Image or torch Tensor.
Parameters: img (PIL Image or Tensor) – Image to be flipped. If img is a Tensor, it is expected to be in […, H, W] format, where … means it can have an arbitrary number of trailing dimensions. Returns: Vertically flipped image. Return type: PIL Image