mmedit.structures.edit_data_sample

Module Contents

Classes

EditDataSample

A data structure interface of MMEditing. They are used as interfaces

Functions

format_label(→ mmengine.structures.LabelData)	Convert label of various python types to mmengine.LabelData.
is_splitable_var(→ bool)	Check whether input is a splitable variable.

mmedit.structures.edit_data_sample.format_label(value: Union[torch.Tensor, numpy.ndarray, Sequence, int], num_classes: int = None) → mmengine.structures.LabelData

Convert label of various python types to mmengine.LabelData.

Supported types are: numpy.ndarray, torch.Tensor, Sequence, int.

参数

• value (torch.Tensor | numpy.ndarray | Sequence | int) – Label value.

• num_classes (int, optional) – The number of classes. If not None, set it to the metainfo. Defaults to None.

返回

The foramtted label data.

返回类型

mmengine.LabelData

mmedit.structures.edit_data_sample.is_splitable_var(var: Any) → bool

Check whether input is a splitable variable.

参数

var (Any) – The input variable to check.

返回

Whether input variable is a splitable variable.

返回类型

bool

class mmedit.structures.edit_data_sample.EditDataSample(*, metainfo: Optional[dict] = None, **kwargs)

Bases: mmengine.structures.BaseDataElement

A data structure interface of MMEditing. They are used as interfaces between different components, e.g., model, visualizer, evaluator, etc. Typically, EditDataSample contains all the information and data from ground-truth and predictions.

EditDataSample inherits from BaseDataElement. See more details in:

https://mmengine.readthedocs.io/en/latest/advanced_tutorials/data_element.html Specifically, an instance of BaseDataElement consists of two components, - metainfo, which contains some meta information,

e.g., img_shape, img_id, etc.

• data, which contains the data used in the loop.

The attributes in EditDataSample are divided into several parts:

• gt_img: Ground truth image(s).

• pred_img: Image(s) of model predictions.

• ref_img: Reference image(s).

• mask: Mask in Inpainting.

• trimap: Trimap in Matting.

• gt_alpha: Ground truth alpha image in Matting.

• pred_alpha: Predicted alpha image in Matting.

• gt_fg: Ground truth foreground image in Matting.

• pred_fg: Predicted foreground image in Matting.

• gt_bg: Ground truth background image in Matting.

• pred_bg: Predicted background image in Matting.

• gt_merged: Ground truth merged image in Matting.

Examples:

 >>> import torch
 >>> import numpy as np
 >>> from mmedit.structures import EditDataSample
 >>> img_meta = dict(img_shape=(800, 1196, 3))
 >>> img = torch.rand((3, 800, 1196))
 >>> data_sample = EditDataSample(gt_img=img, metainfo=img_meta)
 >>> assert 'img_shape' in data_sample.metainfo_keys()
<EditDataSample(

    META INFORMATION
    img_shape: (800, 1196, 3)

    DATA FIELDS
    gt_img: tensor(...)
) at 0x1f6a5a99a00>

property gt_label

This the function to fetch gt label.

返回

gt label.

返回类型

LabelData

META_KEYS

DATA_KEYS

set_predefined_data(data: dict) → None

set or change pre-defined key-value pairs in data_field by parameter data.

参数

data (dict) – A dict contains annotations of image or model predictions.

set_tensor_data(data: dict) → None

convert input data to tensor, and then set or change key-value pairs in data_field by parameter data.

参数

data (dict) – A dict contains annotations of image or model predictions.

set_gt_label(value: Union[numpy.ndarray, torch.Tensor, Sequence[numbers.Number], numbers.Number]) → EditDataSample

Set label of gt_label.

gt_label()

Delete gt label.

classmethod stack(data_samples: Sequence[EditDataSample]) → EditDataSample

Stack a list of data samples to one. All tensor fields will be stacked at first dimension. Otherwise the values will be saved in a list.

参数

data_samples (Sequence['EditDataSample']) – A sequence of EditDataSample to stack.

返回

The stacked data sample.

返回类型

EditDataSample

split(allow_nonseq_value: bool = False) → Sequence[EditDataSample]

Split a sequence of data sample in the first dimension.

参数

allow_nonseq_value (bool) – Whether allow non-sequential data in split operation. If True, non-sequential data will be copied for all split data samples. Otherwise, an error will be raised. Defaults to False.

返回

The list of data samples after splitting.

返回类型

Sequence[EditDataSample]

__len__()

Get the length of the data sample.