Upgrading torchvision 0.8 to 0.9
Unknown0.8.2 → 0.9.1
Fetching 2 releases from GitHub…0 / 2
UnknownUpgrade verdict
DepDelta is still reading the notes for this range, so breaking changes can't be ruled out yet.
In plain terms
- DepDelta is still reading the release notes in this range, so it can't rule out breaking changes yet.
- This page updates as the notes come in.
Must act Still reading release notes
DepDelta is still reading the release notes in this range. This updates as they come in.
Moves torch 1.7.1 → 1.8.1. Your build may pick other versions.
API changes
Generated by DepDelta from published Python source. Not mixed with prose notes.
- added:
export WIDERFace
- removed:
constructor CocoCaptions(root: str, annFile: str, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, transforms: Optional[Callable]=…) -> None - changed:
class CocoCaptions(VisionDataset) → class CocoCaptions(CocoDetection) - changed:
constructor CocoDetection(root: str, annFile: str, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, transforms: Optional[Callable]=…) -> None → constructor CocoDetection(root: str, annFile: str, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, transforms: Optional[Callable]=…)
- added:
static method DatasetFolder.make_dataset(directory: str, class_to_idx: Dict[str, int], extensions: Optional[Tuple[str, ...]]=…, is_valid_file: Optional[Callable[[str], bool]]=…) -> List[Tuple[str, int]]
- added:
field FashionMNIST.mirrors - added:
field KMNIST.mirrors - added:
field MNIST.mirrors
- changed:
function download_url(url: str, root: str, filename: Optional[str]=…, md5: Optional[str]=…) -> None → function download_url(url: str, root: str, filename: Optional[str]=…, md5: Optional[str]=…, max_redirect_hops: int=…) -> None - added:
variable USER_AGENT
- removed:
constructor VOCDetection(root: str, year: str=…, image_set: str=…, download: bool=…, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, transforms: Optional[Callable]=…) - removed:
constructor VOCSegmentation(root: str, year: str=…, image_set: str=…, download: bool=…, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, transforms: Optional[Callable]=…) - removed:
function download_extract(url: str, root: str, filename: str, md5: str) -> None - changed:
class VOCDetection(VisionDataset) → class VOCDetection(_VOCBase) - changed:
class VOCSegmentation(VisionDataset) → class VOCSegmentation(_VOCBase) - added:
property VOCDetection.annotations -> List[str] - added:
property VOCSegmentation.masks -> List[str]
- added:
class WIDERFace(VisionDataset) - added:
constructor WIDERFace(root: str, split: str=…, transform: Optional[Callable]=…, target_transform: Optional[Callable]=…, download: bool=…) -> None - added:
field WIDERFace.ANNOTATIONS_FILE - added:
field WIDERFace.BASE_FOLDER - added:
field WIDERFace.FILE_LIST - added:
method WIDERFace.download() -> None - added:
method WIDERFace.extra_repr() -> str - added:
method WIDERFace.parse_test_annotations_file() -> None - added:
method WIDERFace.parse_train_val_annotations_file() -> None
- added:
export decode_jpeg - added:
export decode_png - added:
export read_file - added:
export write_file
- changed:
function decode_image(input: torch.Tensor) -> torch.Tensor → function decode_image(input: torch.Tensor, mode: ImageReadMode=…) -> torch.Tensor - changed:
function decode_jpeg(input: torch.Tensor) -> torch.Tensor → function decode_jpeg(input: torch.Tensor, mode: ImageReadMode=…) -> torch.Tensor - changed:
function decode_png(input: torch.Tensor) -> torch.Tensor → function decode_png(input: torch.Tensor, mode: ImageReadMode=…) -> torch.Tensor - changed:
function read_image(path: str) -> torch.Tensor → function read_image(path: str, mode: ImageReadMode=…) -> torch.Tensor - added:
class ImageReadMode(Enum) - added:
field ImageReadMode.GRAY - added:
field ImageReadMode.GRAY_ALPHA - added:
field ImageReadMode.RGB - added:
field ImageReadMode.RGB_ALPHA - added:
field ImageReadMode.UNCHANGED
- changed:
function write_video(filename: str, video_array: torch.Tensor, fps: float, video_codec: str=…, options: Optional[Dict[str, Any]]=…) -> None → function write_video(filename: str, video_array: torch.Tensor, fps: float, video_codec: str=…, options: Optional[Dict[str, Any]]=…, audio_array: Optional[torch.Tensor]=…, audio_fps: Optional[float]=…, audio_codec: Optional[str]=…, audio_options: Optional[Dict[str, Any]]=…) -> None
- changed:
constructor AlexNet(num_classes=…) → constructor AlexNet(num_classes: int=…) -> None - changed:
function alexnet(pretrained=…, progress=…, **kwargs) → function alexnet(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> AlexNet - changed:
method AlexNet.forward(x) → method AlexNet.forward(x: torch.Tensor) -> torch.Tensor
- changed:
constructor DenseNet(growth_rate=…, block_config=…, num_init_features=…, bn_size=…, drop_rate=…, num_classes=…, memory_efficient=…) → constructor DenseNet(growth_rate: int=…, block_config: Tuple[int, int, int, int]=…, num_init_features: int=…, bn_size: int=…, drop_rate: float=…, num_classes: int=…, memory_efficient: bool=…) -> None - changed:
function densenet121(pretrained=…, progress=…, **kwargs) → function densenet121(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> DenseNet - changed:
function densenet161(pretrained=…, progress=…, **kwargs) → function densenet161(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> DenseNet - changed:
function densenet169(pretrained=…, progress=…, **kwargs) → function densenet169(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> DenseNet - changed:
function densenet201(pretrained=…, progress=…, **kwargs) → function densenet201(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> DenseNet - changed:
method DenseNet.forward(x) → method DenseNet.forward(x: Tensor) -> Tensor
- changed:
method AnchorGenerator.cached_grid_anchors(grid_sizes, strides) → method AnchorGenerator.cached_grid_anchors(grid_sizes: List[List[int]], strides: List[List[Tensor]]) -> List[Tensor] - changed:
method AnchorGenerator.forward(image_list, feature_maps) → method AnchorGenerator.forward(image_list: ImageList, feature_maps: List[Tensor]) -> List[Tensor] - changed:
method AnchorGenerator.generate_anchors(scales, aspect_ratios, dtype=…, device=…) → method AnchorGenerator.generate_anchors(scales: List[int], aspect_ratios: List[float], dtype: torch.dtype=…, device: torch.device=…) - changed:
method AnchorGenerator.grid_anchors(grid_sizes, strides) → method AnchorGenerator.grid_anchors(grid_sizes: List[List[int]], strides: List[List[Tensor]]) -> List[Tensor] - changed:
method AnchorGenerator.set_cell_anchors(dtype, device) → method AnchorGenerator.set_cell_anchors(dtype: torch.dtype, device: torch.device)
- added:
function mobilenet_backbone(backbone_name, pretrained, fpn, norm_layer=…, trainable_layers=…, returned_layers=…, extra_blocks=…)
- changed:
constructor FasterRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, box_roi_pool=…, box_head=…, box_predictor=…… → constructor FasterRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, rpn_score_thresh=…, box_roi_pool=…, box_hea… - added:
function fasterrcnn_mobilenet_v3_large_320_fpn(pretrained=…, progress=…, num_classes=…, pretrained_backbone=…, trainable_backbone_layers=…, **kwargs) - added:
function fasterrcnn_mobilenet_v3_large_fpn(pretrained=…, progress=…, num_classes=…, pretrained_backbone=…, trainable_backbone_layers=…, **kwargs)
- changed:
constructor ImageList(tensors, image_sizes) → constructor ImageList(tensors: Tensor, image_sizes: List[Tuple[int, int]]) - changed:
method ImageList.to(device) → method ImageList.to(device: torch.device) -> ImageList
- changed:
constructor KeypointRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, box_roi_pool=…, box_head=…, box_predictor… → constructor KeypointRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, rpn_score_thresh=…, box_roi_pool=…, box_h…
- changed:
constructor MaskRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, box_roi_pool=…, box_head=…, box_predictor=…,… → constructor MaskRCNN(backbone, num_classes=…, min_size=…, max_size=…, image_mean=…, image_std=…, rpn_anchor_generator=…, rpn_head=…, rpn_pre_nms_top_n_train=…, rpn_pre_nms_top_n_test=…, rpn_post_nms_top_n_train=…, rpn_post_nms_top_n_test=…, rpn_nms_thresh=…, rpn_fg_iou_thresh=…, rpn_bg_iou_thresh=…, rpn_batch_size_per_image=…, rpn_positive_fraction=…, rpn_score_thresh=…, box_roi_pool=…, box_head=…
- changed:
constructor RetinaNet(backbone, num_classes, min_size=…, max_size=…, image_mean=…, image_std=…, anchor_generator=…, head=…, proposal_matcher=…, score_thresh=…, nms_thresh=…, detections_per_img=…, fg_iou_thresh=…, bg_iou_thresh=…) → constructor RetinaNet(backbone, num_classes, min_size=…, max_size=…, image_mean=…, image_std=…, anchor_generator=…, head=…, proposal_matcher=…, score_thresh=…, nms_thresh=…, detections_per_img=…, fg_iou_thresh=…, bg_iou_thresh=…, topk_candidates=…) - changed:
function retinanet_resnet50_fpn(pretrained=…, progress=…, num_classes=…, pretrained_backbone=…, **kwargs) → function retinanet_resnet50_fpn(pretrained=…, progress=…, num_classes=…, pretrained_backbone=…, trainable_backbone_layers=…, **kwargs)
- changed:
class RoIHeads(torch.nn.Module) → class RoIHeads(nn.Module)
- changed:
constructor RegionProposalNetwork(anchor_generator, head, fg_iou_thresh, bg_iou_thresh, batch_size_per_image, positive_fraction, pre_nms_top_n, post_nms_top_n, nms_thresh) → constructor RegionProposalNetwork(anchor_generator, head, fg_iou_thresh, bg_iou_thresh, batch_size_per_image, positive_fraction, pre_nms_top_n, post_nms_top_n, nms_thresh, score_thresh=…)
- changed:
constructor GoogLeNet(num_classes=…, aux_logits=…, transform_input=…, init_weights=…, blocks=…) → constructor GoogLeNet(num_classes: int=…, aux_logits: bool=…, transform_input: bool=…, init_weights: Optional[bool]=…, blocks: Optional[List[Callable[..., nn.Module]]]=…) -> None - changed:
function googlenet(pretrained=…, progress=…, **kwargs) → function googlenet(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> GoogLeNet - changed:
method GoogLeNet.forward(x) → method GoogLeNet.forward(x: Tensor) -> GoogLeNetOutputs
- changed:
constructor Inception3(num_classes=…, aux_logits=…, transform_input=…, inception_blocks=…, init_weights=…) → constructor Inception3(num_classes: int=…, aux_logits: bool=…, transform_input: bool=…, inception_blocks: Optional[List[Callable[..., nn.Module]]]=…, init_weights: Optional[bool]=…) -> None - changed:
function inception_v3(pretrained=…, progress=…, **kwargs) → function inception_v3(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> Inception3 - changed:
method Inception3.eager_outputs(x: torch.Tensor, aux: Optional[Tensor]) -> InceptionOutputs → method Inception3.eager_outputs(x: Tensor, aux: Optional[Tensor]) -> InceptionOutputs - changed:
method Inception3.forward(x) → method Inception3.forward(x: Tensor) -> InceptionOutputs
- changed:
constructor MNASNet(alpha, num_classes=…, dropout=…) → constructor MNASNet(alpha: float, num_classes: int=…, dropout: float=…) -> None - changed:
function mnasnet0_5(pretrained=…, progress=…, **kwargs) → function mnasnet0_5(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MNASNet - changed:
function mnasnet0_75(pretrained=…, progress=…, **kwargs) → function mnasnet0_75(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MNASNet - changed:
function mnasnet1_0(pretrained=…, progress=…, **kwargs) → function mnasnet1_0(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MNASNet - changed:
function mnasnet1_3(pretrained=…, progress=…, **kwargs) → function mnasnet1_3(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MNASNet - changed:
method MNASNet.forward(x) → method MNASNet.forward(x: Tensor) -> Tensor
- removed:
class MobileNetV2(nn.Module) - removed:
constructor MobileNetV2(num_classes=…, width_mult=…, inverted_residual_setting=…, round_nearest=…, block=…, norm_layer=…) - removed:
function mobilenet_v2(pretrained=…, progress=…, **kwargs) - removed:
method MobileNetV2.forward(x)
- added:
class MobileNetV2(nn.Module) - added:
constructor MobileNetV2(num_classes: int=…, width_mult: float=…, inverted_residual_setting: Optional[List[List[int]]]=…, round_nearest: int=…, block: Optional[Callable[..., nn.Module]]=…, norm_layer: Optional[Callable[..., nn.Module]]=…) -> None - added:
function mobilenet_v2(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MobileNetV2 - added:
method MobileNetV2.forward(x: Tensor) -> Tensor
- added:
class MobileNetV3(nn.Module) - added:
constructor MobileNetV3(inverted_residual_setting: List[InvertedResidualConfig], last_channel: int, num_classes: int=…, block: Optional[Callable[..., nn.Module]]=…, norm_layer: Optional[Callable[..., nn.Module]]=…) -> None - added:
function mobilenet_v3_large(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MobileNetV3 - added:
function mobilenet_v3_small(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> MobileNetV3 - added:
method MobileNetV3.forward(x: Tensor) -> Tensor
- removed:
class QuantizableMobileNetV2(MobileNetV2) - removed:
constructor QuantizableMobileNetV2(*args, **kwargs) - removed:
function mobilenet_v2(pretrained=…, progress=…, quantize=…, **kwargs) - removed:
method QuantizableMobileNetV2.forward(x) - removed:
method QuantizableMobileNetV2.fuse_model()
- added:
class QuantizableMobileNetV2(MobileNetV2) - added:
constructor QuantizableMobileNetV2(*args, **kwargs) - added:
function mobilenet_v2(pretrained=…, progress=…, quantize=…, **kwargs) - added:
method QuantizableMobileNetV2.forward(x) - added:
method QuantizableMobileNetV2.fuse_model()
- added:
class QuantizableMobileNetV3(MobileNetV3) - added:
constructor QuantizableMobileNetV3(*args, **kwargs) - added:
function mobilenet_v3_large(pretrained=…, progress=…, quantize=…, **kwargs) - added:
method QuantizableMobileNetV3.forward(x) - added:
method QuantizableMobileNetV3.fuse_model()
- changed:
constructor ResNet(block, layers, num_classes=…, zero_init_residual=…, groups=…, width_per_group=…, replace_stride_with_dilation=…, norm_layer=…) → constructor ResNet(block: Type[Union[BasicBlock, Bottleneck]], layers: List[int], num_classes: int=…, zero_init_residual: bool=…, groups: int=…, width_per_group: int=…, replace_stride_with_dilation: Optional[List[bool]]=…, norm_layer: Optional[Callable[..., nn.Module]]=…) -> None - changed:
function resnet101(pretrained=…, progress=…, **kwargs) → function resnet101(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnet152(pretrained=…, progress=…, **kwargs) → function resnet152(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnet18(pretrained=…, progress=…, **kwargs) → function resnet18(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnet34(pretrained=…, progress=…, **kwargs) → function resnet34(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnet50(pretrained=…, progress=…, **kwargs) → function resnet50(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnext101_32x8d(pretrained=…, progress=…, **kwargs) → function resnext101_32x8d(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function resnext50_32x4d(pretrained=…, progress=…, **kwargs) → function resnext50_32x4d(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function wide_resnet101_2(pretrained=…, progress=…, **kwargs) → function wide_resnet101_2(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
function wide_resnet50_2(pretrained=…, progress=…, **kwargs) → function wide_resnet50_2(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ResNet - changed:
method ResNet.forward(x) → method ResNet.forward(x: Tensor) -> Tensor
- added:
class LRASPP(nn.Module) - added:
constructor LRASPP(backbone, low_channels, high_channels, num_classes, inter_channels=…) - added:
method LRASPP.forward(input)
- added:
function deeplabv3_mobilenet_v3_large(pretrained=…, progress=…, num_classes=…, aux_loss=…, **kwargs) - added:
function lraspp_mobilenet_v3_large(pretrained=…, progress=…, num_classes=…, **kwargs)
- changed:
constructor ShuffleNetV2(stages_repeats, stages_out_channels, num_classes=…, inverted_residual=…) → constructor ShuffleNetV2(stages_repeats: List[int], stages_out_channels: List[int], num_classes: int=…, inverted_residual: Callable[..., nn.Module]=…) -> None - changed:
function shufflenet_v2_x0_5(pretrained=…, progress=…, **kwargs) → function shufflenet_v2_x0_5(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ShuffleNetV2 - changed:
function shufflenet_v2_x1_0(pretrained=…, progress=…, **kwargs) → function shufflenet_v2_x1_0(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ShuffleNetV2 - changed:
function shufflenet_v2_x1_5(pretrained=…, progress=…, **kwargs) → function shufflenet_v2_x1_5(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ShuffleNetV2 - changed:
function shufflenet_v2_x2_0(pretrained=…, progress=…, **kwargs) → function shufflenet_v2_x2_0(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> ShuffleNetV2 - changed:
method ShuffleNetV2.forward(x) → method ShuffleNetV2.forward(x: Tensor) -> Tensor
- changed:
constructor SqueezeNet(version=…, num_classes=…) → constructor SqueezeNet(version: str=…, num_classes: int=…) -> None - changed:
function squeezenet1_0(pretrained=…, progress=…, **kwargs) → function squeezenet1_0(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> SqueezeNet - changed:
function squeezenet1_1(pretrained=…, progress=…, **kwargs) → function squeezenet1_1(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> SqueezeNet - changed:
method SqueezeNet.forward(x) → method SqueezeNet.forward(x: torch.Tensor) -> torch.Tensor
- changed:
constructor VGG(features, num_classes=…, init_weights=…) → constructor VGG(features: nn.Module, num_classes: int=…, init_weights: bool=…) -> None - changed:
function vgg11(pretrained=…, progress=…, **kwargs) → function vgg11(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg11_bn(pretrained=…, progress=…, **kwargs) → function vgg11_bn(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg13(pretrained=…, progress=…, **kwargs) → function vgg13(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg13_bn(pretrained=…, progress=…, **kwargs) → function vgg13_bn(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg16(pretrained=…, progress=…, **kwargs) → function vgg16(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg16_bn(pretrained=…, progress=…, **kwargs) → function vgg16_bn(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg19(pretrained=…, progress=…, **kwargs) → function vgg19(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
function vgg19_bn(pretrained=…, progress=…, **kwargs) → function vgg19_bn(pretrained: bool=…, progress: bool=…, **kwargs: Any) -> VGG - changed:
method VGG.forward(x) → method VGG.forward(x: torch.Tensor) -> torch.Tensor
- removed:
export _new_empty_tensor
- changed:
function deform_conv2d(input: Tensor, offset: Tensor, weight: Tensor, bias: Optional[Tensor]=…, stride: Tuple[int, int]=…, padding: Tuple[int, int]=…, dilation: Tuple[int, int]=…) -> Tensor → function deform_conv2d(input: Tensor, offset: Tensor, weight: Tensor, bias: Optional[Tensor]=…, stride: Tuple[int, int]=…, padding: Tuple[int, int]=…, dilation: Tuple[int, int]=…, mask: Optional[Tensor]=…) -> Tensor - changed:
method DeformConv2d.forward(input: Tensor, offset: Tensor) -> Tensor → method DeformConv2d.forward(input: Tensor, offset: Tensor, mask: Tensor=…) -> Tensor
- changed:
constructor MultiScaleRoIAlign(featmap_names: List[str], output_size: Union[int, Tuple[int], List[int]], sampling_ratio: int) → constructor MultiScaleRoIAlign(featmap_names: List[str], output_size: Union[int, Tuple[int], List[int]], sampling_ratio: int, *, canonical_scale: int=…, canonical_level: int=…)
- added:
class AutoAugment(torch.nn.Module) - added:
class AutoAugmentPolicy(Enum) - added:
constructor AutoAugment(policy: AutoAugmentPolicy=…, interpolation: InterpolationMode=…, fill: Optional[List[float]]=…) - added:
field AutoAugmentPolicy.CIFAR10 - added:
field AutoAugmentPolicy.IMAGENET - added:
field AutoAugmentPolicy.SVHN - added:
method AutoAugment.forward(img: Tensor) - added:
static method AutoAugment.get_params(transform_num: int) -> Tuple[int, Tensor, Tensor]
- changed:
function affine(img: Tensor, angle: float, translate: List[int], scale: float, shear: List[float], resample: int=…, fillcolor: Optional[int]=…) -> Tensor → function affine(img: Tensor, angle: float, translate: List[int], scale: float, shear: List[float], interpolation: InterpolationMode=…, fill: Optional[List[float]]=…, resample: Optional[int]=…, fillcolor: Optional[List[float]]=…) -> Tensor - changed:
function perspective(img: Tensor, startpoints: List[List[int]], endpoints: List[List[int]], interpolation: int=…, fill: Optional[int]=…) -> Tensor → function perspective(img: Tensor, startpoints: List[List[int]], endpoints: List[List[int]], interpolation: InterpolationMode=…, fill: Optional[List[float]]=…) -> Tensor - changed:
function resize(img: Tensor, size: List[int], interpolation: int=…) -> Tensor → function resize(img: Tensor, size: List[int], interpolation: InterpolationMode=…) -> Tensor - changed:
function resized_crop(img: Tensor, top: int, left: int, height: int, width: int, size: List[int], interpolation: int=…) -> Tensor → function resized_crop(img: Tensor, top: int, left: int, height: int, width: int, size: List[int], interpolation: InterpolationMode=…) -> Tensor - changed:
function rotate(img: Tensor, angle: float, resample: int=…, expand: bool=…, center: Optional[List[int]]=…, fill: Optional[int]=…) -> Tensor → function rotate(img: Tensor, angle: float, interpolation: InterpolationMode=…, expand: bool=…, center: Optional[List[int]]=…, fill: Optional[List[float]]=…, resample: Optional[int]=…) -> Tensor - added:
class InterpolationMode(Enum) - added:
field InterpolationMode.BICUBIC - added:
field InterpolationMode.BILINEAR - added:
field InterpolationMode.BOX - added:
field InterpolationMode.HAMMING - added:
field InterpolationMode.LANCZOS - added:
field InterpolationMode.NEAREST - added:
function adjust_sharpness(img: Tensor, sharpness_factor: float) -> Tensor - added:
function autocontrast(img: Tensor) -> Tensor - added:
function equalize(img: Tensor) -> Tensor - added:
function invert(img: Tensor) -> Tensor - added:
function posterize(img: Tensor, bits: int) -> Tensor - added:
function solarize(img: Tensor, threshold: float) -> Tensor - added:
variable pil_modes_mapping
- changed:
function affine(img, matrix, resample=…, fillcolor=…) → function affine(img, matrix, interpolation=…, fill=…) - changed:
function rotate(img, angle, resample=…, expand=…, center=…, fill=…) → function rotate(img, angle, interpolation=…, expand=…, center=…, fill=…) - added:
function adjust_sharpness(img, sharpness_factor) - added:
function autocontrast(img) - added:
function equalize(img) - added:
function invert(img) - added:
function posterize(img, bits) - added:
function solarize(img, threshold)
- changed:
function affine(img: Tensor, matrix: List[float], resample: int=…, fillcolor: Optional[int]=…) -> Tensor → function affine(img: Tensor, matrix: List[float], interpolation: str=…, fill: Optional[List[float]]=…) -> Tensor - changed:
function perspective(img: Tensor, perspective_coeffs: List[float], interpolation: int=…, fill: Optional[int]=…) -> Tensor → function perspective(img: Tensor, perspective_coeffs: List[float], interpolation: str=…, fill: Optional[List[float]]=…) -> Tensor - changed:
function resize(img: Tensor, size: List[int], interpolation: int=…) -> Tensor → function resize(img: Tensor, size: List[int], interpolation: str=…) -> Tensor - changed:
function rotate(img: Tensor, matrix: List[float], resample: int=…, expand: bool=…, fill: Optional[int]=…) -> Tensor → function rotate(img: Tensor, matrix: List[float], interpolation: str=…, expand: bool=…, fill: Optional[List[float]]=…) -> Tensor - added:
function adjust_sharpness(img: Tensor, sharpness_factor: float) -> Tensor - added:
function autocontrast(img: Tensor) -> Tensor - added:
function equalize(img: Tensor) -> Tensor - added:
function invert(img: Tensor) -> Tensor - added:
function posterize(img: Tensor, bits: int) -> Tensor - added:
function solarize(img: Tensor, threshold: float) -> Tensor
- changed:
constructor RandomAffine(degrees, translate=…, scale=…, shear=…, resample=…, fillcolor=…) → constructor RandomAffine(degrees, translate=…, scale=…, shear=…, interpolation=…, fill=…, fillcolor=…, resample=…) - changed:
constructor RandomRotation(degrees, resample=…, expand=…, center=…, fill=…) → constructor RandomRotation(degrees, interpolation=…, expand=…, center=…, fill=…, resample=…) - changed:
static method ColorJitter.get_params(brightness, contrast, saturation, hue) → static method ColorJitter.get_params(brightness: Optional[List[float]], contrast: Optional[List[float]], saturation: Optional[List[float]], hue: Optional[List[float]]) -> Tuple[Tensor, Optional[float], Optional[float], Optional[float], Optional[float]] - added:
class RandomAdjustSharpness(torch.nn.Module) - added:
class RandomAutocontrast(torch.nn.Module) - added:
class RandomEqualize(torch.nn.Module) - added:
class RandomInvert(torch.nn.Module) - added:
class RandomPosterize(torch.nn.Module) - added:
class RandomSolarize(torch.nn.Module) - added:
constructor RandomAdjustSharpness(sharpness_factor, p=…) - added:
constructor RandomAutocontrast(p=…) - added:
constructor RandomEqualize(p=…) - added:
constructor RandomInvert(p=…) - added:
constructor RandomPosterize(bits, p=…) - added:
constructor RandomSolarize(threshold, p=…) - added:
export InterpolationMode - added:
method RandomAdjustSharpness.forward(img) - added:
method RandomAutocontrast.forward(img) - added:
method RandomEqualize.forward(img) - added:
method RandomInvert.forward(img) - added:
method RandomPosterize.forward(img) - added:
method RandomSolarize.forward(img)
- removed:
variable irange - changed:
function make_grid(tensor: Union[torch.Tensor, List[torch.Tensor]], nrow: int=…, padding: int=…, normalize: bool=…, range: Optional[Tuple[int, int]]=…, scale_each: bool=…, pad_value: int=…) -> torch.Tensor → function make_grid(tensor: Union[torch.Tensor, List[torch.Tensor]], nrow: int=…, padding: int=…, normalize: bool=…, value_range: Optional[Tuple[int, int]]=…, scale_each: bool=…, pad_value: int=…, **kwargs) -> torch.Tensor - changed:
function save_image(tensor: Union[torch.Tensor, List[torch.Tensor]], fp: Union[Text, pathlib.Path, BinaryIO], nrow: int=…, padding: int=…, normalize: bool=…, range: Optional[Tuple[int, int]]=…, scale_each: bool=…, pad_value: int=…, format: Optional[str]=…) -> None → function save_image(tensor: Union[torch.Tensor, List[torch.Tensor]], fp: Union[Text, pathlib.Path, BinaryIO], format: Optional[str]=…, **kwargs) -> None - added:
function draw_bounding_boxes(image: torch.Tensor, boxes: torch.Tensor, labels: Optional[List[str]]=…, colors: Optional[List[Union[str, Tuple[int, int, int]]]]=…, fill: Optional[bool]=…, width: int=…, font: Optional[str]=…, font_size: int=…) -> torch.Tensor