UnAV-100
A benchmark dataset for evaluating omni-modal temporal grounding capability, focusing on event time localization in audio-visual content.
Timeline (last 90 days)
OmniGround reaches 37.5 mIoU on the LongVALE benchmark and 35.7 mIoU on the UnAV-100 benchmark, both achieving SOTA levels for omni-modal temporal grounding.
LongVALE focuses on joint vision-audio-language understanding, while UnAV-100 focuses on audio event localization; both are commonly used benchmarks for audio-visual temporal grounding.
All Facts (2)
OmniGround reaches 37.5 mIoU on the LongVALE benchmark and 35.7 mIoU on the UnAV-100 benchmark, both achieving SOTA levels for omni-modal temporal grounding.
50%UnverifiedLongVALE focuses on joint vision-audio-language understanding, while UnAV-100 focuses on audio event localization; both are commonly used benchmarks for audio-visual temporal grounding.
50%