vis4d.eval.scalabel¶

Scalabel evaluator.

class ScalabelDetectEvaluator(annotation_path, config=None, mask_threshold=0.0)[source]¶

Scalabel 2D detection evaluation class.

Initialize the evaluator.

__repr__()[source]¶

Concise representation of the dataset evaluator.

Return type:: str

evaluate(metric)[source]¶

Evaluate the dataset.

Return type:: tuple[Dict[str, Union[float, int, Tensor]], str]

process_batch(frame_ids, sample_names, sequence_names, pred_boxes, pred_classes, pred_scores, pred_masks=None)[source]¶

Process tracking results.

Return type:: None

property metrics: list[str]¶: Supported metrics.

class ScalabelEvaluator(annotation_path, config=None)[source]¶

Scalabel base evaluation class.

Initialize the evaluator.

evaluate(metric)[source]¶

Evaluate the dataset.

Return type:: tuple[Dict[str, Union[float, int, Tensor]], str]

gather(gather_func)[source]¶

Gather variables in case of distributed setting (if needed).

Parameters:: gather_func (Callable[[Any], Any]) – Gather function.
Return type:: None

process_batch(*args, **kwargs)[source]¶

Process sample and update confusion matrix.

Return type:: None

reset()[source]¶

Reset the evaluator.

Return type:: None

class ScalabelTrackEvaluator(annotation_path, config=None, mask_threshold=0.0)[source]¶

Scalabel 2D tracking evaluation class.

Initialize the evaluator.

__repr__()[source]¶

Concise representation of the dataset evaluator.

Return type:: str

evaluate(metric)[source]¶

Evaluate the dataset.

Return type:: tuple[Dict[str, Union[float, int, Tensor]], str]

process_batch(frame_ids, sample_names, sequence_names, pred_boxes, pred_classes, pred_scores, pred_track_ids, pred_masks=None)[source]¶

Process tracking results.

Return type:: None

property metrics: list[str]¶: Supported metrics.

Modules

`vis4d.eval.scalabel.base`	Scalabel base evaluator.
`vis4d.eval.scalabel.detect`	Scalabel detection evaluator.
`vis4d.eval.scalabel.track`	Scalabel tracking evaluator.