AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Class-Aware Sounding Objects Localization via Audiovisual Correspondence

Di Hu, Yake Wei, Rui Qian, Weiyao Lin, Ruihua Song, Ji-Rong Wen

TPAMI 2021 · Detection and segmentation › Open-set and few-shot detection · approach: Object-level representation learning · 47 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 silent-area suppression (Prod) (ablation)Ours · MUSIC-SyntheticCIoU@0.3: 2.6 -> 32.3Improved
add silent-area suppression (Prod) (ablation)Ours · VGGSound-SyntheticCIoU@0.3: 3.1 -> 13.4Improved
add audiovisual consistency loss L_c (ablation)Ours · MUSIC-SyntheticCIoU@0.3: 18.0 -> 32.3Improved
add audiovisual consistency loss L_c (ablation)Ours · Realistic DailyLifeCIoU@0.3: 20.1 -> 22.6Improved
increase clusters from 100 to 5000 (ablation)Ours · VGGSound-SyntheticCIoU@0.3: 13.4 -> 0.8Hurt
use 98 clusters (true category number) instead of 100 (ablation)Ours · VGGSound (single source)IoU@0.5: 49.5 -> 50.1No clear effect
remove 3 categories from first-stage data (ablation)Ours · MUSIC-SyntheticCIoU@0.3: 32.3 -> 8.4Hurt
remove 3 categories from first-stage data (ablation)Ours · MUSIC-soloIoU@0.5: 51.4 -> 46.7Hurt
add 15% multi-source noise to first-stage videos (ablation)Ours · MUSIC-SyntheticCIoU@0.3: 32.3 -> 19.3Hurt
add alternative localization-classification optimization (ablation)Ours · VGGSound, DailyLife, MUSIClocalizationImproved
use ground-truth object labels (Oracle-v) instead of pseudo categoriesOurs · VGGSound-SyntheticCIoU@0.3: 13.4 -> 21.4Improved
use ground-truth audio labels (Oracle-a) instead of pseudo categoriesOurs · VGGSound-SyntheticCIoU@0.3: 13.4 -> 13.7No clear effect

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