AI Paper Atlas

Research map / Video / Action recognition

Learnable Irrelevant Modality Dropout for Multimodal Action Recognition on Modality-Specific Annotated Videos

Saghir Alfasly, Jian Lu, Chen Xu, Yuru Zou

CVPR 2022 · Video › Action recognition · approach: Disentangled action recognition · 31 citations (OpenAlex)

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
Use learnable relevance-based audio masking and multimodal fusionIMD action recognition · Kinetics400Top-1: 80.9 -> 82.3Improved
Use learnable relevance-based audio masking and multimodal fusionIMD-B · UCF-101Top-1: 98.9 -> 99.1Improved
Use learnable relevance-based audio masking and multimodal fusionIMD-B · HMDB51Top-1: 70.2 -> 71.3Improved
Use learnable relevance-based audio masking and multimodal fusionIMD-B · Kinetics-SoundsTop-1: 85.0 -> 90.48Improved
Use BERT rather than word2vec or GloVe for SAVLD label embeddings (ablation)SAVLD · Kinetics400Top-1: 80.9 -> 82.8; 81.2 -> 82.8Improved
Use cosine rather than Euclidean similarity for BERT label matching (ablation)SAVLD · Kinetics400Top-1: 82.8 -> 82.8No clear effect
Add audio and flow to RGB (ablation)Proposed multimodal framework · Kinetics400Top-1: 79.1 -> 82.8Improved
Use IOU rather than Dice overlap loss (ablation)IMD · Kinetics400Top-1: 81.6 -> 81.7No clear effect
Set the K-Y_a ratio to 10-20 (ablation)IMD relevance ground truth · Kinetics400Top-1: 81.7 -> 82.3Improved
Set the relevance threshold alpha to 0.25 rather than 0.75 (ablation)IMD · Kinetics400Top-1: 80.9 -> 82.8Improved
Mask irrelevant audio instead of weighting audio at every relevance score (ablation)IMD · Kinetics400Top-1: 80.4 -> 82.8Improved
Add intra-class cross-modality augmentation (ablation)IMD · Kinetics400Top-1: 82.3 -> 82.8Improved

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