MaskFormer
Last updated
Last updated
This is a recently introduced model so the API hasn’t been tested extensively. There may be some bugs or slight breaking changes to fix it in the future. If you see something strange, file a Github Issue.
The MaskFormer model was proposed in Per-Pixel Classification is Not All You Need for Semantic Segmentation by Bowen Cheng, Alexander G. Schwing, Alexander Kirillov. MaskFormer addresses semantic segmentation with a mask classification paradigm instead of performing classic pixel-level classification.
The abstract from the paper is the following:
Modern approaches typically formulate semantic segmentation as a per-pixel classification task, while instance-level segmentation is handled with an alternative mask classification. Our key insight: mask classification is sufficiently general to solve both semantic- and instance-level segmentation tasks in a unified manner using the exact same model, loss, and training procedure. Following this observation, we propose MaskFormer, a simple mask classification model which predicts a set of binary masks, each associated with a single global class label prediction. Overall, the proposed mask classification-based method simplifies the landscape of effective approaches to semantic and panoptic segmentation tasks and shows excellent empirical results. In particular, we observe that MaskFormer outperforms per-pixel classification baselines when the number of classes is large. Our mask classification-based method outperforms both current state-of-the-art semantic (55.6 mIoU on ADE20K) and panoptic segmentation (52.7 PQ on COCO) models.
Tips:
MaskFormer’s Transformer decoder is identical to the decoder of DETR. During training, the authors of DETR did find it helpful to use auxiliary losses in the decoder, especially to help the model output the correct number of objects of each class. If you set the parameter use_auxilary_loss
of MaskFormerConfig to True
, then prediction feedforward neural networks and Hungarian losses are added after each decoder layer (with the FFNs sharing parameters).
If you want to train the model in a distributed environment across multiple nodes, then one should update the get_num_masks
function inside in the MaskFormerLoss
class of modeling_maskformer.py
. When training on multiple nodes, this should be set to the average number of target masks across all nodes, as can be seen in the original implementation here.
One can use MaskFormerImageProcessor to prepare images for the model and optional targets for the model.
To get the final segmentation, depending on the task, you can call post_process_semantic_segmentation() or post_process_panoptic_segmentation(). Both tasks can be solved using MaskFormerForInstanceSegmentation output, panoptic segmentation accepts an optional label_ids_to_fuse
argument to fuse instances of the target object/s (e.g. sky) together.
The figure below illustrates the architecture of MaskFormer. Taken from the original paper.
This model was contributed by francesco. The original code can be found here.
Image Segmentation
All notebooks that illustrate inference as well as fine-tuning on custom data with MaskFormer can be found here.
( encoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Nonepixel_decoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Nonetransformer_decoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Noneencoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonepixel_decoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonetransformer_decoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonehidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Noneattentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None )
Parameters
encoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the encoder model (backbone).
pixel_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the pixel decoder model (FPN).
transformer_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, sequence_length, hidden_size)
) — Last hidden states (final feature map) of the last stage of the transformer decoder model.
encoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the encoder model at the output of each stage.
pixel_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the pixel decoder model at the output of each stage.
transformer_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, sequence_length, hidden_size)
. Hidden-states (also called feature maps) of the transformer decoder at the output of each stage.
hidden_states tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
containing encoder_hidden_states
, pixel_decoder_hidden_states
and decoder_hidden_states
attentions (tuple(torch.FloatTensor)
, optional, returned when output_attentions=True
is passed or when config.output_attentions=True
) — Tuple of torch.FloatTensor
(one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length)
. Attentions weights from Detr’s decoder after the attention softmax, used to compute the weighted average in the self-attention heads.
Class for outputs of MaskFormerModel. This class returns all the needed hidden states to compute the logits.
( loss: typing.Optional[torch.FloatTensor] = Noneclass_queries_logits: FloatTensor = Nonemasks_queries_logits: FloatTensor = Noneauxiliary_logits: FloatTensor = Noneencoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Nonepixel_decoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Nonetransformer_decoder_last_hidden_state: typing.Optional[torch.FloatTensor] = Noneencoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonepixel_decoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonetransformer_decoder_hidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Nonehidden_states: typing.Optional[typing.Tuple[torch.FloatTensor]] = Noneattentions: typing.Optional[typing.Tuple[torch.FloatTensor]] = None )
Parameters
loss (torch.Tensor
, optional) — The computed loss, returned when labels are present.
class_queries_logits (torch.FloatTensor
) — A tensor of shape (batch_size, num_queries, num_labels + 1)
representing the proposed classes for each query. Note the + 1
is needed because we incorporate the null class.
masks_queries_logits (torch.FloatTensor
) — A tensor of shape (batch_size, num_queries, height, width)
representing the proposed masks for each query.
encoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the encoder model (backbone).
pixel_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the pixel decoder model (FPN).
transformer_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, sequence_length, hidden_size)
) — Last hidden states (final feature map) of the last stage of the transformer decoder model.
encoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the encoder model at the output of each stage.
pixel_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the pixel decoder model at the output of each stage.
transformer_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, sequence_length, hidden_size)
. Hidden-states of the transformer decoder at the output of each stage.
hidden_states tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
containing encoder_hidden_states
, pixel_decoder_hidden_states
and decoder_hidden_states
.
attentions (tuple(torch.FloatTensor)
, optional, returned when output_attentions=True
is passed or when config.output_attentions=True
) — Tuple of torch.FloatTensor
(one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length)
. Attentions weights from Detr’s decoder after the attention softmax, used to compute the weighted average in the self-attention heads.
Class for outputs of MaskFormerForInstanceSegmentation.
This output can be directly passed to post_process_semantic_segmentation() or or post_process_instance_segmentation() or post_process_panoptic_segmentation() depending on the task. Please, see [`~MaskFormerImageProcessor] for details regarding usage.
( fpn_feature_size: int = 256mask_feature_size: int = 256no_object_weight: float = 0.1use_auxiliary_loss: bool = Falsebackbone_config: typing.Optional[typing.Dict] = Nonedecoder_config: typing.Optional[typing.Dict] = Noneinit_std: float = 0.02init_xavier_std: float = 1.0dice_weight: float = 1.0cross_entropy_weight: float = 1.0mask_weight: float = 20.0output_auxiliary_logits: typing.Optional[bool] = None**kwargs )
Parameters
mask_feature_size (int
, optional, defaults to 256) — The masks’ features size, this value will also be used to specify the Feature Pyramid Network features’ size.
no_object_weight (float
, optional, defaults to 0.1) — Weight to apply to the null (no object) class.
use_auxiliary_loss(bool
, optional, defaults to False
) — If True
MaskFormerForInstanceSegmentationOutput
will contain the auxiliary losses computed using the logits from each decoder’s stage.
backbone_config (Dict
, optional) — The configuration passed to the backbone, if unset, the configuration corresponding to swin-base-patch4-window12-384
will be used.
decoder_config (Dict
, optional) — The configuration passed to the transformer decoder model, if unset the base config for detr-resnet-50
will be used.
init_std (float
, optional, defaults to 0.02) — The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
init_xavier_std (float
, optional, defaults to 1) — The scaling factor used for the Xavier initialization gain in the HM Attention map module.
dice_weight (float
, optional, defaults to 1.0) — The weight for the dice loss.
cross_entropy_weight (float
, optional, defaults to 1.0) — The weight for the cross entropy loss.
mask_weight (float
, optional, defaults to 20.0) — The weight for the mask loss.
output_auxiliary_logits (bool
, optional) — Should the model output its auxiliary_logits
or not.
Raises
ValueError
ValueError
— Raised if the backbone model type selected is not in ["swin"]
or the decoder model type selected is not in ["detr"]
This is the configuration class to store the configuration of a MaskFormerModel. It is used to instantiate a MaskFormer model according to the specified arguments, defining the model architecture. Instantiating a configuration with the defaults will yield a similar configuration to that of the MaskFormer facebook/maskformer-swin-base-ade architecture trained on ADE20k-150.
Configuration objects inherit from PretrainedConfig and can be used to control the model outputs. Read the documentation from PretrainedConfig for more information.
Currently, MaskFormer only supports the Swin Transformer as backbone.
Examples:
Copied
from_backbone_and_decoder_configs
( backbone_config: PretrainedConfigdecoder_config: PretrainedConfig**kwargs ) → MaskFormerConfig
Parameters
backbone_config (PretrainedConfig) — The backbone configuration.
decoder_config (PretrainedConfig) — The transformer decoder configuration to use.
Returns
An instance of a configuration object
Instantiate a MaskFormerConfig (or a derived class) from a pre-trained backbone model configuration and DETR model configuration.
( do_resize: bool = Truesize: typing.Dict[str, int] = Nonesize_divisor: int = 32resample: Resampling = <Resampling.BILINEAR: 2>do_rescale: bool = Truerescale_factor: float = 0.00392156862745098do_normalize: bool = Trueimage_mean: typing.Union[float, typing.List[float]] = Noneimage_std: typing.Union[float, typing.List[float]] = Noneignore_index: typing.Optional[int] = Nonedo_reduce_labels: bool = False**kwargs )
Parameters
do_resize (bool
, optional, defaults to True
) — Whether to resize the input to a certain size
.
size (int
, optional, defaults to 800) — Resize the input to the given size. Only has an effect if do_resize
is set to True
. If size is a sequence like (width, height)
, 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)
.
max_size (int
, optional, defaults to 1333) — The largest size an image dimension can have (otherwise it’s capped). Only has an effect if do_resize
is set to True
.
resample (int
, optional, defaults to PIL.Image.Resampling.BILINEAR
) — An optional resampling filter. This can be one of PIL.Image.Resampling.NEAREST
, PIL.Image.Resampling.BOX
, PIL.Image.Resampling.BILINEAR
, PIL.Image.Resampling.HAMMING
, PIL.Image.Resampling.BICUBIC
or PIL.Image.Resampling.LANCZOS
. Only has an effect if do_resize
is set to True
.
size_divisor (int
, optional, defaults to 32) — Some backbones need images divisible by a certain number. If not passed, it defaults to the value used in Swin Transformer.
do_rescale (bool
, optional, defaults to True
) — Whether to rescale the input to a certain scale
.
rescale_factor (float
, optional, defaults to 1/ 255) — Rescale the input by the given factor. Only has an effect if do_rescale
is set to True
.
do_normalize (bool
, optional, defaults to True
) — Whether or not to normalize the input with mean and standard deviation.
image_mean (int
, optional, defaults to [0.485, 0.456, 0.406]
) — The sequence of means for each channel, to be used when normalizing images. Defaults to the ImageNet mean.
image_std (int
, optional, defaults to [0.229, 0.224, 0.225]
) — The sequence of standard deviations for each channel, to be used when normalizing images. Defaults to the ImageNet std.
ignore_index (int
, optional) — Label to be assigned to background pixels in segmentation maps. If provided, segmentation map pixels denoted with 0 (background) will be replaced with ignore_index
.
do_reduce_labels (bool
, optional, defaults to False
) — Whether or not to decrement all label values of segmentation maps by 1. Usually used for datasets where 0 is used for background, and background itself is not included in all classes of a dataset (e.g. ADE20k). The background label will be replaced by ignore_index
.
Constructs a MaskFormer image processor. The image processor can be used to prepare image(s) and optional targets for the model.
This image processor inherits from BaseImageProcessor
which contains most of the main methods. Users should refer to this superclass for more information regarding those methods.
preprocess
( images: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]]segmentation_maps: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')], NoneType] = Noneinstance_id_to_semantic_id: typing.Union[typing.Dict[int, int], NoneType] = Nonedo_resize: typing.Optional[bool] = Nonesize: typing.Union[typing.Dict[str, int], NoneType] = Nonesize_divisor: typing.Optional[int] = Noneresample: Resampling = Nonedo_rescale: typing.Optional[bool] = Nonerescale_factor: typing.Optional[float] = Nonedo_normalize: typing.Optional[bool] = Noneimage_mean: typing.Union[float, typing.List[float], NoneType] = Noneimage_std: typing.Union[float, typing.List[float], NoneType] = Noneignore_index: typing.Optional[int] = Nonedo_reduce_labels: typing.Optional[bool] = Nonereturn_tensors: typing.Union[str, transformers.utils.generic.TensorType, NoneType] = Nonedata_format: typing.Union[str, transformers.image_utils.ChannelDimension] = <ChannelDimension.FIRST: 'channels_first'>input_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None**kwargs )
encode_inputs
( pixel_values_list: typing.List[typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]]]segmentation_maps: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]] = Noneinstance_id_to_semantic_id: typing.Union[typing.List[typing.Dict[int, int]], typing.Dict[int, int], NoneType] = Noneignore_index: typing.Optional[int] = Nonereduce_labels: bool = Falsereturn_tensors: typing.Union[str, transformers.utils.generic.TensorType, NoneType] = Noneinput_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None ) → BatchFeature
Parameters
pixel_values_list (List[ImageInput]
) — List of images (pixel values) to be padded. Each image should be a tensor of shape (channels, height, width)
.
segmentation_maps (ImageInput
, optional) — The corresponding semantic segmentation maps with the pixel-wise annotations.
(bool
, optional, defaults to True
): Whether or not to pad images up to the largest image in a batch and create a pixel mask.
If left to the default, will return a pixel mask that is:
1 for pixels that are real (i.e. not masked),
0 for pixels that are padding (i.e. masked).
instance_id_to_semantic_id (List[Dict[int, int]]
or Dict[int, int]
, optional) — A mapping between object instance ids and class ids. If passed, segmentation_maps
is treated as an instance segmentation map where each pixel represents an instance id. Can be provided as a single dictionary with a global/dataset-level mapping or as a list of dictionaries (one per image), to map instance ids in each image separately.
return_tensors (str
or TensorType, optional) — If set, will return tensors instead of NumPy arrays. If set to 'pt'
, return PyTorch torch.Tensor
objects.
Returns
A BatchFeature with the following fields:
pixel_values — Pixel values to be fed to a model.
pixel_mask — Pixel mask to be fed to a model (when =True
or if pixel_mask
is in self.model_input_names
).
mask_labels — Optional list of mask labels of shape (labels, height, width)
to be fed to a model (when annotations
are provided).
class_labels — Optional list of class labels of shape (labels)
to be fed to a model (when annotations
are provided). They identify the labels of mask_labels
, e.g. the label of mask_labels[i][j]
if class_labels[i][j]
.
Pad images up to the largest image in a batch and create a corresponding pixel_mask
.
MaskFormer addresses semantic segmentation with a mask classification paradigm, thus input segmentation maps will be converted to lists of binary masks and their respective labels. Let’s see an example, assuming segmentation_maps = [[2,6,7,9]]
, the output will contain mask_labels = [[1,0,0,0],[0,1,0,0],[0,0,1,0],[0,0,0,1]]
(four binary masks) and class_labels = [2,6,7,9]
, the labels for each mask.
post_process_semantic_segmentation
( outputstarget_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = None ) → List[torch.Tensor]
Parameters
outputs (MaskFormerForInstanceSegmentation) — Raw outputs of the model.
target_sizes (List[Tuple[int, int]]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction. If left to None, predictions will not be resized.
Returns
List[torch.Tensor]
A list of length batch_size
, where each item is a semantic segmentation map of shape (height, width) corresponding to the target_sizes entry (if target_sizes
is specified). Each entry of each torch.Tensor
correspond to a semantic class id.
Converts the output of MaskFormerForInstanceSegmentation into semantic segmentation maps. Only supports PyTorch.
post_process_instance_segmentation
( outputsthreshold: float = 0.5mask_threshold: float = 0.5overlap_mask_area_threshold: float = 0.8target_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = Nonereturn_coco_annotation: typing.Optional[bool] = Falsereturn_binary_maps: typing.Optional[bool] = False ) → List[Dict]
Parameters
outputs (MaskFormerForInstanceSegmentation) — Raw outputs of the model.
threshold (float
, optional, defaults to 0.5) — The probability score threshold to keep predicted instance masks.
mask_threshold (float
, optional, defaults to 0.5) — Threshold to use when turning the predicted masks into binary values.
overlap_mask_area_threshold (float
, optional, defaults to 0.8) — The overlap mask area threshold to merge or discard small disconnected parts within each binary instance mask.
target_sizes (List[Tuple]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction. If left to None, predictions will not be resized.
return_coco_annotation (bool
, optional, defaults to False
) — If set to True
, segmentation maps are returned in COCO run-length encoding (RLE) format.
return_binary_maps (bool
, optional, defaults to False
) — If set to True
, segmentation maps are returned as a concatenated tensor of binary segmentation maps (one per detected instance).
Returns
List[Dict]
A list of dictionaries, one per image, each dictionary containing two keys:
segmentation — A tensor of shape (height, width)
where each pixel represents a segment_id
or List[List]
run-length encoding (RLE) of the segmentation map if return_coco_annotation is set to True
. Set to None
if no mask if found above threshold
.
segments_info — A dictionary that contains additional information on each segment.
id — An integer representing the segment_id
.
label_id — An integer representing the label / semantic class id corresponding to segment_id
.
score — Prediction score of segment with segment_id
.
Converts the output of MaskFormerForInstanceSegmentationOutput
into instance segmentation predictions. Only supports PyTorch.
post_process_panoptic_segmentation
( outputsthreshold: float = 0.5mask_threshold: float = 0.5overlap_mask_area_threshold: float = 0.8label_ids_to_fuse: typing.Optional[typing.Set[int]] = Nonetarget_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = None ) → List[Dict]
Parameters
outputs (MaskFormerForInstanceSegmentationOutput
) — The outputs from MaskFormerForInstanceSegmentation.
threshold (float
, optional, defaults to 0.5) — The probability score threshold to keep predicted instance masks.
mask_threshold (float
, optional, defaults to 0.5) — Threshold to use when turning the predicted masks into binary values.
overlap_mask_area_threshold (float
, optional, defaults to 0.8) — The overlap mask area threshold to merge or discard small disconnected parts within each binary instance mask.
label_ids_to_fuse (Set[int]
, optional) — The labels in this state will have all their instances be fused together. For instance we could say there can only be one sky in an image, but several persons, so the label ID for sky would be in that set, but not the one for person.
target_sizes (List[Tuple]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction in batch. If left to None, predictions will not be resized.
Returns
List[Dict]
A list of dictionaries, one per image, each dictionary containing two keys:
segmentation — a tensor of shape (height, width)
where each pixel represents a segment_id
, set to None
if no mask if found above threshold
. If target_sizes
is specified, segmentation is resized to the corresponding target_sizes
entry.
segments_info — A dictionary that contains additional information on each segment.
id — an integer representing the segment_id
.
label_id — An integer representing the label / semantic class id corresponding to segment_id
.
was_fused — a boolean, True
if label_id
was in label_ids_to_fuse
, False
otherwise. Multiple instances of the same class / label were fused and assigned a single segment_id
.
score — Prediction score of segment with segment_id
.
Converts the output of MaskFormerForInstanceSegmentationOutput
into image panoptic segmentation predictions. Only supports PyTorch.
( *args**kwargs )
__call__
( imagessegmentation_maps = None**kwargs )
encode_inputs
( pixel_values_list: typing.List[typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]]]segmentation_maps: typing.Union[ForwardRef('PIL.Image.Image'), numpy.ndarray, ForwardRef('torch.Tensor'), typing.List[ForwardRef('PIL.Image.Image')], typing.List[numpy.ndarray], typing.List[ForwardRef('torch.Tensor')]] = Noneinstance_id_to_semantic_id: typing.Union[typing.List[typing.Dict[int, int]], typing.Dict[int, int], NoneType] = Noneignore_index: typing.Optional[int] = Nonereduce_labels: bool = Falsereturn_tensors: typing.Union[str, transformers.utils.generic.TensorType, NoneType] = Noneinput_data_format: typing.Union[str, transformers.image_utils.ChannelDimension, NoneType] = None ) → BatchFeature
Parameters
pixel_values_list (List[ImageInput]
) — List of images (pixel values) to be padded. Each image should be a tensor of shape (channels, height, width)
.
segmentation_maps (ImageInput
, optional) — The corresponding semantic segmentation maps with the pixel-wise annotations.
(bool
, optional, defaults to True
): Whether or not to pad images up to the largest image in a batch and create a pixel mask.
If left to the default, will return a pixel mask that is:
1 for pixels that are real (i.e. not masked),
0 for pixels that are padding (i.e. masked).
instance_id_to_semantic_id (List[Dict[int, int]]
or Dict[int, int]
, optional) — A mapping between object instance ids and class ids. If passed, segmentation_maps
is treated as an instance segmentation map where each pixel represents an instance id. Can be provided as a single dictionary with a global/dataset-level mapping or as a list of dictionaries (one per image), to map instance ids in each image separately.
return_tensors (str
or TensorType, optional) — If set, will return tensors instead of NumPy arrays. If set to 'pt'
, return PyTorch torch.Tensor
objects.
Returns
A BatchFeature with the following fields:
pixel_values — Pixel values to be fed to a model.
pixel_mask — Pixel mask to be fed to a model (when =True
or if pixel_mask
is in self.model_input_names
).
mask_labels — Optional list of mask labels of shape (labels, height, width)
to be fed to a model (when annotations
are provided).
class_labels — Optional list of class labels of shape (labels)
to be fed to a model (when annotations
are provided). They identify the labels of mask_labels
, e.g. the label of mask_labels[i][j]
if class_labels[i][j]
.
Pad images up to the largest image in a batch and create a corresponding pixel_mask
.
MaskFormer addresses semantic segmentation with a mask classification paradigm, thus input segmentation maps will be converted to lists of binary masks and their respective labels. Let’s see an example, assuming segmentation_maps = [[2,6,7,9]]
, the output will contain mask_labels = [[1,0,0,0],[0,1,0,0],[0,0,1,0],[0,0,0,1]]
(four binary masks) and class_labels = [2,6,7,9]
, the labels for each mask.
post_process_semantic_segmentation
( outputstarget_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = None ) → List[torch.Tensor]
Parameters
outputs (MaskFormerForInstanceSegmentation) — Raw outputs of the model.
target_sizes (List[Tuple[int, int]]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction. If left to None, predictions will not be resized.
Returns
List[torch.Tensor]
A list of length batch_size
, where each item is a semantic segmentation map of shape (height, width) corresponding to the target_sizes entry (if target_sizes
is specified). Each entry of each torch.Tensor
correspond to a semantic class id.
Converts the output of MaskFormerForInstanceSegmentation into semantic segmentation maps. Only supports PyTorch.
post_process_instance_segmentation
( outputsthreshold: float = 0.5mask_threshold: float = 0.5overlap_mask_area_threshold: float = 0.8target_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = Nonereturn_coco_annotation: typing.Optional[bool] = Falsereturn_binary_maps: typing.Optional[bool] = False ) → List[Dict]
Parameters
outputs (MaskFormerForInstanceSegmentation) — Raw outputs of the model.
threshold (float
, optional, defaults to 0.5) — The probability score threshold to keep predicted instance masks.
mask_threshold (float
, optional, defaults to 0.5) — Threshold to use when turning the predicted masks into binary values.
overlap_mask_area_threshold (float
, optional, defaults to 0.8) — The overlap mask area threshold to merge or discard small disconnected parts within each binary instance mask.
target_sizes (List[Tuple]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction. If left to None, predictions will not be resized.
return_coco_annotation (bool
, optional, defaults to False
) — If set to True
, segmentation maps are returned in COCO run-length encoding (RLE) format.
return_binary_maps (bool
, optional, defaults to False
) — If set to True
, segmentation maps are returned as a concatenated tensor of binary segmentation maps (one per detected instance).
Returns
List[Dict]
A list of dictionaries, one per image, each dictionary containing two keys:
segmentation — A tensor of shape (height, width)
where each pixel represents a segment_id
or List[List]
run-length encoding (RLE) of the segmentation map if return_coco_annotation is set to True
. Set to None
if no mask if found above threshold
.
segments_info — A dictionary that contains additional information on each segment.
id — An integer representing the segment_id
.
label_id — An integer representing the label / semantic class id corresponding to segment_id
.
score — Prediction score of segment with segment_id
.
Converts the output of MaskFormerForInstanceSegmentationOutput
into instance segmentation predictions. Only supports PyTorch.
post_process_panoptic_segmentation
( outputsthreshold: float = 0.5mask_threshold: float = 0.5overlap_mask_area_threshold: float = 0.8label_ids_to_fuse: typing.Optional[typing.Set[int]] = Nonetarget_sizes: typing.Union[typing.List[typing.Tuple[int, int]], NoneType] = None ) → List[Dict]
Parameters
outputs (MaskFormerForInstanceSegmentationOutput
) — The outputs from MaskFormerForInstanceSegmentation.
threshold (float
, optional, defaults to 0.5) — The probability score threshold to keep predicted instance masks.
mask_threshold (float
, optional, defaults to 0.5) — Threshold to use when turning the predicted masks into binary values.
overlap_mask_area_threshold (float
, optional, defaults to 0.8) — The overlap mask area threshold to merge or discard small disconnected parts within each binary instance mask.
label_ids_to_fuse (Set[int]
, optional) — The labels in this state will have all their instances be fused together. For instance we could say there can only be one sky in an image, but several persons, so the label ID for sky would be in that set, but not the one for person.
target_sizes (List[Tuple]
, optional) — List of length (batch_size), where each list item (Tuple[int, int]]
) corresponds to the requested final size (height, width) of each prediction in batch. If left to None, predictions will not be resized.
Returns
List[Dict]
A list of dictionaries, one per image, each dictionary containing two keys:
segmentation — a tensor of shape (height, width)
where each pixel represents a segment_id
, set to None
if no mask if found above threshold
. If target_sizes
is specified, segmentation is resized to the corresponding target_sizes
entry.
segments_info — A dictionary that contains additional information on each segment.
id — an integer representing the segment_id
.
label_id — An integer representing the label / semantic class id corresponding to segment_id
.
was_fused — a boolean, True
if label_id
was in label_ids_to_fuse
, False
otherwise. Multiple instances of the same class / label were fused and assigned a single segment_id
.
score — Prediction score of segment with segment_id
.
Converts the output of MaskFormerForInstanceSegmentationOutput
into image panoptic segmentation predictions. Only supports PyTorch.
( config: MaskFormerConfig )
Parameters
config (MaskFormerConfig) — Model configuration class with all the parameters of the model. Initializing with a config file does not load the weights associated with the model, only the configuration. Check out the from_pretrained() method to load the model weights.
The bare MaskFormer Model outputting raw hidden-states without any specific head on top. This model is a PyTorch torch.nn.Module sub-class. Use it as a regular PyTorch Module and refer to the PyTorch documentation for all matter related to general usage and behavior.
forward
( pixel_values: Tensorpixel_mask: typing.Optional[torch.Tensor] = Noneoutput_hidden_states: typing.Optional[bool] = Noneoutput_attentions: typing.Optional[bool] = Nonereturn_dict: typing.Optional[bool] = None ) → transformers.models.maskformer.modeling_maskformer.MaskFormerModelOutput or tuple(torch.FloatTensor)
Parameters
pixel_values (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Pixel values. Pixel values can be obtained using AutoImageProcessor. See MaskFormerImageProcessor.call() for details.
pixel_mask (torch.LongTensor
of shape (batch_size, height, width)
, optional) — Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]
:
1 for pixels that are real (i.e. not masked),
0 for pixels that are padding (i.e. masked).
output_hidden_states (bool
, optional) — Whether or not to return the hidden states of all layers. See hidden_states
under returned tensors for more detail.
output_attentions (bool
, optional) — Whether or not to return the attentions tensors of Detr’s decoder attention layers.
return_dict (bool
, optional) — Whether or not to return a ~MaskFormerModelOutput
instead of a plain tuple.
Returns
transformers.models.maskformer.modeling_maskformer.MaskFormerModelOutput or tuple(torch.FloatTensor)
A transformers.models.maskformer.modeling_maskformer.MaskFormerModelOutput or a tuple of torch.FloatTensor
(if return_dict=False
is passed or when config.return_dict=False
) comprising various elements depending on the configuration (MaskFormerConfig) and inputs.
encoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the encoder model (backbone).
pixel_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the pixel decoder model (FPN).
transformer_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, sequence_length, hidden_size)
) — Last hidden states (final feature map) of the last stage of the transformer decoder model.
encoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the encoder model at the output of each stage.
pixel_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the pixel decoder model at the output of each stage.
transformer_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, sequence_length, hidden_size)
. Hidden-states (also called feature maps) of the transformer decoder at the output of each stage.
hidden_states tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
containing encoder_hidden_states
, pixel_decoder_hidden_states
and decoder_hidden_states
attentions (tuple(torch.FloatTensor)
, optional, returned when output_attentions=True
is passed or when config.output_attentions=True
) — Tuple of torch.FloatTensor
(one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length)
. Attentions weights from Detr’s decoder after the attention softmax, used to compute the weighted average in the self-attention heads.
The MaskFormerModel forward method, overrides the __call__
special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.
Examples:
Copied
( config: MaskFormerConfig )
forward
( pixel_values: Tensormask_labels: typing.Optional[typing.List[torch.Tensor]] = Noneclass_labels: typing.Optional[typing.List[torch.Tensor]] = Nonepixel_mask: typing.Optional[torch.Tensor] = Noneoutput_auxiliary_logits: typing.Optional[bool] = Noneoutput_hidden_states: typing.Optional[bool] = Noneoutput_attentions: typing.Optional[bool] = Nonereturn_dict: typing.Optional[bool] = None ) → transformers.models.maskformer.modeling_maskformer.MaskFormerForInstanceSegmentationOutput or tuple(torch.FloatTensor)
Parameters
pixel_values (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Pixel values. Pixel values can be obtained using AutoImageProcessor. See MaskFormerImageProcessor.call() for details.
pixel_mask (torch.LongTensor
of shape (batch_size, height, width)
, optional) — Mask to avoid performing attention on padding pixel values. Mask values selected in [0, 1]
:
1 for pixels that are real (i.e. not masked),
0 for pixels that are padding (i.e. masked).
output_hidden_states (bool
, optional) — Whether or not to return the hidden states of all layers. See hidden_states
under returned tensors for more detail.
output_attentions (bool
, optional) — Whether or not to return the attentions tensors of Detr’s decoder attention layers.
return_dict (bool
, optional) — Whether or not to return a ~MaskFormerModelOutput
instead of a plain tuple.
mask_labels (List[torch.Tensor]
, optional) — List of mask labels of shape (num_labels, height, width)
to be fed to a model
class_labels (List[torch.LongTensor]
, optional) — list of target class labels of shape (num_labels, height, width)
to be fed to a model. They identify the labels of mask_labels
, e.g. the label of mask_labels[i][j]
if class_labels[i][j]
.
Returns
transformers.models.maskformer.modeling_maskformer.MaskFormerForInstanceSegmentationOutput or tuple(torch.FloatTensor)
A transformers.models.maskformer.modeling_maskformer.MaskFormerForInstanceSegmentationOutput or a tuple of torch.FloatTensor
(if return_dict=False
is passed or when config.return_dict=False
) comprising various elements depending on the configuration (MaskFormerConfig) and inputs.
loss (torch.Tensor
, optional) — The computed loss, returned when labels are present.
class_queries_logits (torch.FloatTensor
) — A tensor of shape (batch_size, num_queries, num_labels + 1)
representing the proposed classes for each query. Note the + 1
is needed because we incorporate the null class.
masks_queries_logits (torch.FloatTensor
) — A tensor of shape (batch_size, num_queries, height, width)
representing the proposed masks for each query.
encoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the encoder model (backbone).
pixel_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, num_channels, height, width)
) — Last hidden states (final feature map) of the last stage of the pixel decoder model (FPN).
transformer_decoder_last_hidden_state (torch.FloatTensor
of shape (batch_size, sequence_length, hidden_size)
) — Last hidden states (final feature map) of the last stage of the transformer decoder model.
encoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the encoder model at the output of each stage.
pixel_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, num_channels, height, width)
. Hidden-states (also called feature maps) of the pixel decoder model at the output of each stage.
transformer_decoder_hidden_states (tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
(one for the output of the embeddings + one for the output of each stage) of shape (batch_size, sequence_length, hidden_size)
. Hidden-states of the transformer decoder at the output of each stage.
hidden_states tuple(torch.FloatTensor)
, optional, returned when output_hidden_states=True
is passed or when config.output_hidden_states=True
) — Tuple of torch.FloatTensor
containing encoder_hidden_states
, pixel_decoder_hidden_states
and decoder_hidden_states
.
attentions (tuple(torch.FloatTensor)
, optional, returned when output_attentions=True
is passed or when config.output_attentions=True
) — Tuple of torch.FloatTensor
(one for each layer) of shape (batch_size, num_heads, sequence_length, sequence_length)
. Attentions weights from Detr’s decoder after the attention softmax, used to compute the weighted average in the self-attention heads.
The MaskFormerForInstanceSegmentation forward method, overrides the __call__
special method.
Although the recipe for forward pass needs to be defined within this function, one should call the Module
instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them.
Examples:
Semantic segmentation example:
Copied
Panoptic segmentation example:
Copied