Skip to content

unitorch.cli.models.swin¤

SwinProcessor¤

Tip

core/process/swin is the section for configuration of SwinProcessor.

Bases: SwinProcessor

Swin Transformer processor for image tasks.

Source code in src/unitorch/cli/models/swin/processing.py 
20
21
22
23
24
25
26
def __init__(
    self,
    vision_config_path: str,
):
    super().__init__(
        vision_config_path=vision_config_path,
    )

from_config classmethod ¤

from_config(config, **kwargs)
Source code in src/unitorch/cli/models/swin/processing.py 
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
@classmethod
@config_defaults_init("core/process/swin")
def from_config(cls, config, **kwargs):
    config.set_default_section("core/process/swin")
    pretrained_name = config.getoption(
        "pretrained_name", "swin-tiny-patch4-window7-224"
    )
    vision_config_path = config.getoption("vision_config_path", None)
    vision_config_path = pop_value(
        vision_config_path,
        nested_dict_value(pretrained_swin_infos, pretrained_name, "vision_config"),
    )

    vision_config_path = cached_path(vision_config_path)

    return {
        "vision_config_path": vision_config_path,
    }

_image_classification ¤

_image_classification(image: Union[Image, str])
Source code in src/unitorch/cli/models/swin/processing.py 
47
48
49
50
51
52
53
54
55
@register_process("core/process/swin/image_classification")
def _image_classification(
    self,
    image: Union[Image.Image, str],
):
    if isinstance(image, str):
        image = Image.open(image)
    outputs = super().classification(image=image)
    return TensorInputs(pixel_values=outputs.pixel_values)

SwinForImageClassification¤

Tip

core/model/classification/swin is the section for configuration of SwinForImageClassification.

Bases: SwinForImageClassification

Swin Transformer model for image classification.

Source code in src/unitorch/cli/models/swin/modeling.py 
24
25
26
27
28
29
30
31
32
def __init__(
    self,
    config_path: str,
    num_classes: Optional[int] = 1,
):
    super().__init__(
        config_path=config_path,
        num_classes=num_classes,
    )

from_config classmethod ¤

from_config(config, **kwargs)
Source code in src/unitorch/cli/models/swin/modeling.py 
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
@classmethod
@config_defaults_init("core/model/classification/swin")
def from_config(cls, config, **kwargs):
    config.set_default_section("core/model/classification/swin")
    pretrained_name = config.getoption(
        "pretrained_name", "swin-tiny-patch4-window7-224"
    )
    config_path = config.getoption("config_path", None)
    config_path = pop_value(
        config_path,
        nested_dict_value(pretrained_swin_infos, pretrained_name, "config"),
    )

    config_path = cached_path(config_path)
    num_classes = config.getoption("num_classes", 1)

    inst = cls(
        config_path=config_path,
        num_classes=num_classes,
    )
    pretrained_weight_path = config.getoption("pretrained_weight_path", None)
    weight_path = pop_value(
        pretrained_weight_path,
        nested_dict_value(pretrained_swin_infos, pretrained_name, "weight"),
        check_none=False,
    )
    if weight_path is not None:
        inst.from_pretrained(weight_path)

    return inst

forward ¤

forward(pixel_values: Tensor)
Source code in src/unitorch/cli/models/swin/modeling.py 
65
66
67
68
69
70
71
@autocast(device_type=("cuda" if torch.cuda.is_available() else "cpu"))
def forward(
    self,
    pixel_values: torch.Tensor,
):
    outputs = super().forward(pixel_values=pixel_values)
    return ClassificationOutputs(outputs=outputs)