List your company

Active Speaker Localization on EasyCom (test)

86.32mAPOurs AV(cor+eng+box)
42.556853.918465.2876.6416Jan 6, 2022

Evaluation results

MethodLinks
Ours AV(cor+eng+box)Audio Channels=4, Modality=Audio-Visual, Features=Cross-correlation, Energy, Bounding box2022.0186.32
Ours AV(cor+box)Audio Channels=4, Modality=Audio-Visual, Features=Cross-correlation, Bounding box2022.0186.25
Ours AV(spec)Audio Channels=4, Modality=Audio-Visual, Features=Spectrogram2022.0185.49
Ours AV(cor)Audio Channels=4, Modality=Audio-Visual, Features=Cross-correlation2022.0184.14
Ours AV(cor+eng)Audio Channels=4, Modality=Audio-Visual, Features=Cross-correlation, Energy2022.0183.32
Ours AV(spec)-2chAudio Channels=2, Modality=Audio-Visual, Features=Spectrogram2022.0183.3
Ours AV(cor)-2chAudio Channels=2, Modality=Audio-Visual, Features=Cross-correlation2022.0180
AV-rawaudioAudio Channels=4, Modality=Audio-Visual2022.0172.32
TalkNet (AVA)Training Data=AVA, Audio Channels=1, Modality=Audio-Visual2022.0169.13
MRC (EasyCom)Training Data=EasyCom, Modality=Video-only2022.0164.24
Ours AV(eng)Audio Channels=4, Modality=Audio-Visual, Features=Energy2022.0162.68
BinauralAVLocAudio Channels=2, Modality=Audio-Visual2022.0160.75
DOA+imageAudio Channels=4, Modality=Audio-Visual2022.0154.27
DOA+headboxAudio Channels=4, Modality=Audio-Visual2022.0152.62
MRC (AVA)Training Data=AVA, Modality=Video-only2022.0146.6
TalkNet (EasyCom)Training Data=EasyCom, Audio Channels=1, Modality=Audio-Visual2022.0144.24