images

Concrete ImageRepresentation subclasses for common detectors.

class graphnet.models.data_representation.images.images.IC86Image(*args, **kwargs)[source]

Bases: ImageRepresentation

IceCube-86 images (main array + optional DeepCore tensors).

Construct IC86Image.

Parameters:

  • pixel_definition (NodeDefinition) – Pulse → DOM row features (NodeDefinition).

  • input_feature_names (List[str]) – Raw input column names.

  • include_lower_dc (bool, default: True) – Include lower DeepCore grid.

  • include_upper_dc (bool, default: True) – Include upper DeepCore grid.

  • string_label (str, default: 'string') – DOM string column in pixel rows.

  • dom_number_label (str, default: 'dom_number') – DOM index column in pixel rows.

  • dtype (Optional[dtype], default: torch.float32) – Tensor dtype for images.

  • detector (Optional[Detector], default: None) – IceCube86; default standardizes all but coordinates.

  • args (Any)

  • kwargs (Any)

Return type:

object

class graphnet.models.data_representation.images.images.ExamplePrometheusImage(*args, **kwargs)[source]

Bases: ImageRepresentation

Example Prometheus-style single-image layout (tutorial scripts only).

Construct ExamplePrometheusImage.

Parameters:

  • pixel_definition (NodeDefinition) – Pulse → sensor row features (NodeDefinition).

  • input_feature_names (List[str]) – Raw input column names.

  • string_label (str, default: 'sensor_string_id') – String id column in pixel rows.

  • dom_number_label (str, default: 'sensor_id') – Sensor id column (internal grid key name).

  • dtype (Optional[dtype], default: torch.float32) – Tensor dtype for images.

  • detector (Optional[Detector], default: None) – ORCA150 by default.

  • args (Any)

  • kwargs (Any)

Return type:

object