AI Paper Atlas

Research map / Vision-language and multimodal / Cross-modal learning

Contrastive Positive Sample Propagation Along the Audio-Visual Event Line

Jinxing Zhou, Dan Guo, Meng Wang

TPAMI 2022 · Vision-language and multimodal › Cross-modal learning · approach: Audio-visual source localization · 91 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add the PSP module (ablation)localization network · AVE (fully supervised)accuracy (%): 73.7 -> 77.8Improved
add the PSP module (ablation)localization network · AVE (weakly supervised)accuracy (%): 70.2 -> 73.5Improved
prune weak and negative connections instead of keeping all (tau) (ablation)PSP · AVE (fully supervised)accuracy (%): 75.9 -> 77.8Improved
prune weak and negative connections instead of keeping all (tau) (ablation)PSP · AVE (weakly supervised)accuracy (%): 71.2 -> 73.5Improved
also prune weak connections, not only negative ones (ablation)PSP · AVE (fully supervised)accuracy (%): 76.0 -> 77.8Improved
add self-attention before the Bi-LSTMs (ablation)PSP · AVE (fully supervised)accuracy (%): 77.8 -> 75.4Hurt
add self-attention before the Bi-LSTMs (ablation)PSP · AVE (weakly supervised)accuracy (%): 73.5 -> 70.8Hurt
add L_avpsp pair-level loss (ablation)PSP · AVE (fully supervised)accuracy (%): 76.6 -> 77.8Improved
add L_avpsp pair-level loss (ablation)AVEL · AVE (fully supervised)accuracy (%): 69.8 -> 71.3Improved
add weighting branch (extra FC) (ablation)PSP · AVE (weakly supervised)accuracy (%): 71.6 -> 73.5Improved
add weighting branch (extra FC) (ablation)AVEL · AVE (weakly supervised)accuracy (%): 66.9 -> 69.2Improved
add segment-level PSA_S contrastive loss (ablation)PSP · VGGSound-AVEL100k (fully supervised)accuracy (%): 58.3 -> 59.6Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy