UWAV generates temporally coherent, uncertainty-weighted segment-level pseudo-labels with a transformer pre-trained on a larger dataset, then uses them with mixup and class rebalancing to train an audio-visual video parser, improving state-of-the-art results on the LLP and AVE benchmarks.
Visual scene graphs for audio source separation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
UWAV: Uncertainty-weighted Weakly-supervised Audio-Visual Video Parsing
UWAV generates temporally coherent, uncertainty-weighted segment-level pseudo-labels with a transformer pre-trained on a larger dataset, then uses them with mixup and class rebalancing to train an audio-visual video parser, improving state-of-the-art results on the LLP and AVE benchmarks.