`mmedit.structures`¶

Package Contents¶

Classes¶

`EditDataSample`	A data structure interface of MMEditing. They are used as interfaces
`PixelData`	Data structure for pixel-level annnotations or predictions.

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

Bases: mmengine.structures.BaseDataElement

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

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, PixelData
 >>> data_sample = EditDataSample()
 >>> img_meta = dict(img_shape=(800, 1196, 3))
 >>> img = torch.rand((3, 800, 1196))
 >>> gt_img = PixelData(data=img, metainfo=img_meta)
 >>> data_sample.gt_img = gt_img
 >>> assert 'img_shape' in data_sample.gt_img.metainfo_keys()
<EditDataSample(

    META INFORMATION

    DATA FIELDS
    _gt_img: <PixelData(

            META INFORMATION
            img_shape: (800, 1196, 3)

            DATA FIELDS
            data: tensor([[[0.8069, 0.4279,  ..., 0.6603, 0.0292],

                        ...,

                        [0.8139, 0.0908,  ..., 0.4964, 0.9672]]])
        ) at 0x1f6ae000af0>
    gt_img: <PixelData(

            META INFORMATION
            img_shape: (800, 1196, 3)

            DATA FIELDS
            data: tensor([[[0.8069, 0.4279,  ..., 0.6603, 0.0292],

                        ...,

                        [0.8139, 0.0908,  ..., 0.4964, 0.9672]]])
        ) at 0x1f6ae000af0>
) at 0x1f6a5a99a00>

property gt_img: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_img in PixelData.

返回: data element
返回类型: PixelData

property gt_samples: EditDataSample¶

This is the function to fetch gt_samples.

返回: gt samples.
返回类型: EditDataSample

property noise: torch.Tensor¶

This is the function to fetch noise.

返回: noise.
返回类型: torch.Tensor

property pred_img: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch pred_img in PixelData.

返回: data element
返回类型: PixelData

property fake_img: Union[mmedit.structures.pixel_data.PixelData, torch.Tensor]¶

This is the function to fetch fake_img.

返回: The fake img.
返回类型: Union[PixelData, Tensor]

property img_lq: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch img_lq in PixelData.

返回: data element
返回类型: PixelData

property ref_img: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch ref_img.

返回: data element
返回类型: PixelData

property ref_lq: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch ref_lq.

返回: data element
返回类型: PixelData

property gt_unsharp: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_unsharp in PixelData.

返回: data element
返回类型: PixelData

property mask: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch mask.

返回: data element
返回类型: PixelData

property gt_heatmap: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_heatmap.

返回: data element
返回类型: PixelData

property pred_heatmap: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch pred_heatmap.

返回: data element
返回类型: PixelData

property trimap: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch trimap.

返回: data element
返回类型: PixelData

property gt_alpha: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_alpha.

返回: data element
返回类型: PixelData

property pred_alpha: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch pred_alpha.

返回: data element
返回类型: PixelData

property gt_fg: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_fg.

返回: data element
返回类型: PixelData

property pred_fg: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch pred_fg.

返回: _description_
返回类型: PixelData

property gt_bg: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_bg.

返回: data element
返回类型: PixelData

property pred_bg: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch pred_bg in PixelData.

返回: data element
返回类型: PixelData

property gt_merged: mmedit.structures.pixel_data.PixelData¶

This is the function to fetch gt_merged in PixelData.

返回: _description_
返回类型: PixelData

property sample_model: str¶

This is the function to fetch sample model.

返回: Mode of Sample model.
返回类型: str

property ema: EditDataSample¶

This is the function to fetch ema results.

返回: Results of the ema model.
返回类型: EditDataSample

property orig: EditDataSample¶

This is the function to fetch original results.

返回: Results of the ema model.
返回类型: EditDataSample

property gt_label¶

This the function to fetch gt label.

返回: gt label.
返回类型: LabelData

gt_img()¶: This is the function to fetch gt_img.

gt_samples()¶: This is the function to delete gt_samples.

noise()¶: This is the functionto delete noise.

pred_img()¶: This is the function to fetch pred_img.

fake_img()¶: This is the function to delete fake_img.

img_lq()¶: This is the function to delete img_lq.

ref_img()¶: This is the function to fetch ref_img.

ref_lq()¶: This is the function to delete ref_lq.

gt_unsharp()¶: This is the function to delete gt_unsharp.

mask()¶: This is the function to delete mask.

gt_heatmap()¶: This is the function to delete gt_heatmap.

pred_heatmap()¶: This is the function to fetch pred_heatmap.

trimap()¶: This is the function to delete trimap.

gt_alpha()¶: This is the function to delete gt_alpha.

pred_alpha()¶: This is the function to delete pred_alpha.

gt_fg()¶: This is the function to delete gt_fg.

pred_fg()¶: This is the function to delete pred_fg.

gt_bg()¶: This is the function to delete gt_bg.

pred_bg()¶: This is the function to fetch pred_bg.

gt_merged()¶: This is the function to fetch gt_merged.

sample_model()¶: This is the function to delete sample model.

ema()¶: This is the function to delete ema results.

orig()¶: This is the function to delete ema results.

set_gt_label(value: Union[numpy.ndarray, torch.Tensor, Sequence[numbers.Number], numbers.Number]) → EditDataSample¶: Set label of gt_label.

gt_label()¶: This is the function to delete gt label.

class mmedit.structures.PixelData(*, metainfo: Optional[dict] = None, **kwargs)[源代码]¶

Bases: mmengine.structures.PixelData

Data structure for pixel-level annnotations or predictions.

Different from parent class:: Support value.ndim == 4 for frames.

All data items in data_fields of PixelData meet the following requirements:

They all have 3 dimensions in orders of channel, height, and width.
They should have the same height and width.

实际案例

>>> metainfo = dict(
...     img_id=random.randint(0, 100),
...     img_shape=(random.randint(400, 600), random.randint(400, 600)))
>>> image = np.random.randint(0, 255, (4, 20, 40))
>>> featmap = torch.randint(0, 255, (10, 20, 40))
>>> pixel_data = PixelData(metainfo=metainfo,
...                        image=image,
...                        featmap=featmap)
>>> print(pixel_data)
>>> (20, 40)

>>> # slice
>>> slice_data = pixel_data[10:20, 20:40]
>>> assert slice_data.shape == (10, 10)
>>> slice_data = pixel_data[10, 20]
>>> assert slice_data.shape == (1, 1)

__setattr__(name: str, value: Union[torch.Tensor, numpy.ndarray])¶

Set attributes of PixelData.

If the dimension of value is 2 and its shape meet the demand, it will automatically expend its channel-dimension.

参数

name (str) – The key to access the value, stored in PixelData.
value (Union[torch.Tensor, np.ndarray]) – The value to store in. The type of value must be torch.Tensor or np.ndarray, and its shape must meet the requirements of PixelData.

mmedit.structures¶

Package Contents¶

Classes¶

`mmedit.structures`¶