Skip to content

rslearn.train.transforms

transforms

rslearn transforms.

Sequential

Bases: Module

Sequentially apply provided transforms.

Each transform should accept (input_dict, target_dict) and return an updated tuple.

Source code in rslearn/train/transforms/__init__.py 
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
class Sequential(torch.nn.Module):
    """Sequentially apply provided transforms.

    Each transform should accept (input_dict, target_dict) and return an updated
    tuple.
    """

    def __init__(self, *args: Any) -> None:
        """Initialize a new Sequential from a list of transforms."""
        super().__init__()
        self.transforms = torch.nn.ModuleList(args)

    def forward(
        self, input_dict: dict[str, Any], target_dict: dict[str, Any]
    ) -> tuple[dict[str, Any], dict[str, Any]]:
        """Apply each specified transform."""
        for transform in self.transforms:
            input_dict, target_dict = transform(input_dict, target_dict)
        return input_dict, target_dict

forward 

forward(input_dict: dict[str, Any], target_dict: dict[str, Any]) -> tuple[dict[str, Any], dict[str, Any]]

Apply each specified transform.

Source code in rslearn/train/transforms/__init__.py 
20
21
22
23
24
25
26
def forward(
    self, input_dict: dict[str, Any], target_dict: dict[str, Any]
) -> tuple[dict[str, Any], dict[str, Any]]:
    """Apply each specified transform."""
    for transform in self.transforms:
        input_dict, target_dict = transform(input_dict, target_dict)
    return input_dict, target_dict