search

pyannote / pyannote-metrics

A toolkit for reproducible evaluation, diagnostic, and error analysis of speaker diarization systems
http://pyannote.github.io/pyannote-metrics
MIT License
186 stars 33 forks source link

Collar removal is applied twice in DiarizationErrorRate #16

Closed hbredin closed 7 years ago

hbredin commented 7 years ago

Description

Because of this and this, collar removal is applied twice in DiarizationErrorRate.compute_components()

Steps/Code to Reproduce

>>> from pyannote.core import Annotation, Segment
>>> from pyannote.metrics.diarization import DiarizationErrorRate
>>> der = DiarizationErrorRate(collar=1)
>>> reference = Annotation()
>>> reference[Segment(0, 10)] = 'A'
>>> hypothesis = Annotation()
>>> der(reference, hypothesis, detailed=True)['total']

Expected Results

9.0

Actual Results

8.0

Versions

pyannote.algorithms==0.7.3 pyannote.core==1.0.4 pyannote.metrics==1.2 pyannote.parser==0.6.5