nerfbaselines.results

class nerfbaselines.results.DatasetInfo[source]

Bases: TypedDict

default_metric: str
description: str
id: str
licenses: NotRequired[List[LicenseSpec]]
metrics: List[MetricInfo]
name: str
paper_authors: List[str]
paper_title: str
scenes: List[SceneInfo]
class nerfbaselines.results.MetricInfo[source]

Bases: TypedDict

ascending: bool
description: str
id: str
name: str
class nerfbaselines.results.SceneInfo[source]

Bases: TypedDict

id: str
name: str
nerfbaselines.results.compile_dataset_results(results_path: Path | str, dataset: str, scenes: List[str] | None = None) → Dict[str, Any][source]

Compile the results.json file from the results repository.

nerfbaselines.results.format_duration(seconds: float | None) → str[source]
nerfbaselines.results.format_memory(memory: float | None) → str[source]
nerfbaselines.results.get_dataset_info(dataset: str) → DatasetInfo[source]

Get the dataset info from the dataset repository.

Parameters:

dataset – The dataset name (type).

Returns:

The dataset info.

nerfbaselines.results.get_method_info_from_spec(spec: MethodSpec) → MethodInfo[source]

Get the method info from the method spec without importing the method dependencies. If the method info is loaded from the method implementation, the method spec should have the metadata fields.

Parameters:

spec – The method spec.

Returns:

The method info.

nerfbaselines.results.load_metrics_from_results(results: Dict) → Dict[str, List[float]][source]

Load the metrics from a results file (obtained from evaluation).

Parameters:

results – A dictionary of results.

Returns:

A dictionary containing the metrics.

nerfbaselines.results.render_markdown_dataset_results_table(results, method_links: Literal['paper', 'website', 'results', 'none'] = 'none') → str[source]

Generates a markdown table from the output of the compile_dataset_results method.

Parameters:

results – Output of the nerfbaselines.results.compile_dataset_results method.