Multi-sound source localization on MUSIC-Duet (test) benchmark leaderboard · SOTA2 Research

Multi-sound source localization on MUSIC-Duet (test)

82.7CIoU@0.3Generation-Analysis-Refinement

Evaluation Results

Method
Generation-Analysis-Refinement2026.04	82.7	56.7	—	53.2
Generation-Analysis-Refinement2026.04	80.8	54.7	—	51.4
Qwen2.5-Omni2026.04	50.6	47.2	—	40.8
OA-SSL2026.04	45.9	61.4	—	36.1
T-VSL2024.04	43.2	62.9	—	35.9
AVGN2024.04	39.3	59.2	—	32.4
DSOL2024.04	38.8	—	—	31.3
Proposed Method2024.03	38.6	52.1	72.5	30.1
NoPrior2026.04	38.6	52.1	—	30.1
Alignment2024.04	38.3	57.8	—	30.9
Mix-and-Localize2024.04	38.1	57.4	—	30.7
FNAC2024.04	37.9	57.3	—	30.6
AudioCLIP2024.04	36.9	56.1	—	29.2
MarginNCE2024.04	36.1	55.1	—	28.5
EZ-VSL2024.04	34.3	—	—	26.7
LVS2024.04	33.1	—	—	26.9
AVGN2024.03	32.5	50.6	57.2	24.6
AVGN2026.04	32.5	50.6	—	24.6
Wav2CLIP2024.04	32.2	52.6	—	27.2
CoarsetoFine2024.04	31.8	—	—	24.1
DSOL2024.03	30.1	—	—	22.3
OTS2024.04	29.4	51.2	—	24.3
MiniCPM-o2026.04	27.7	29.3	—	23.6
Mix-and-Localize2024.03	26.5	47.5	54.1	21.5
Mix-and-Localize2026.04	26.5	47.5	—	21.5
EZ-VSL2024.03	24.3	—	—	21.3
EZ-VSL2026.04	24.3	—	—	21.3
LVS2024.03	22.5	—	—	20.9
Attention10k2024.03	21.6	—	—	19.6
Attention 10k2026.04	21.6	—	—	19.6
InteractiveOmni2026.04	20	28.8	—	17
CoarseToFIne2024.03	17.6	—	—	20.6
CoarseToFine2026.04	17.6	—	—	20.6
DMC2024.03	17.5	—	—	21.1
DMC2026.04	17.5	—	—	21.1
OTS2024.03	13.3	11.6	17.7	18.5
OTS2026.04	13.3	11.6	—	18.5