type

benchmarks.hota.PanopticMetricRunner

class PanopticMetricRunner:

Stream (gt, pred) panoptic maps into HOTA/CLEAR/Identity and aggregate.

Surfaces a flat ``{metric_name: float}`` of the headline scores: HOTA/DetA/ AssA/LocA (``___AUC`` aggregates) plus MOTA and IDF1.

Members

method

  • __init__
  • computeAggregate the streamed sequences into headline scalar scores.
  • end_sequenceClose the current sequence on every metric.
  • start_sequenceOpen a new sequence of ``length`` frames on every metric.
  • updateFeed one frame's ground-truth and predicted panoptic maps.

Source: unitrack/benchmarks/hota/metric.py:8