Research map / Vision-language and multimodal / Cross-modal learning
Listen to Look into the Future: Audio-Visual Egocentric Gaze Anticipation
Bolin Lai, Fiona Ryan, Wenqi Jia, Miao Liu, James M Rehg
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add audio via spatial fusion module (ablation) | MViT gaze anticipation · Ego4D | F1 score: 37.2 -> 38.6 | Improved |
| add audio via temporal fusion module (ablation) | MViT gaze anticipation · Aria | F1 score: 57.5 -> 58.6 | Improved |
| combine spatial and temporal modules (STS) instead of one (ablation) | CSTS · Ego4D | F1 score: 38.7 -> 39.2 | Improved |
| use spatial-temporal separable fusion instead of joint self-attention (ablation) | CSTS · Ego4D | F1 score: 38.5 -> 39.2 | Improved |
| use STS instead of bilinear joint fusion (ablation) | CSTS · Aria | F1 score: 57.7 -> 59.3 | Improved |
| add post-fusion contrastive loss to STS (ablation) | CSTS · Ego4D | F1 score: 39.2 -> 39.7 | Improved |
| apply contrastive loss on raw embeddings (vanilla) in STS (ablation) | CSTS · Ego4D | F1 score: 39.2 -> 39.0 | Hurt |
| apply vanilla contrastive loss to vanilla self-attention (ablation) | Vanilla SA · Ego4D | F1 score: 38.5 -> 38.5 | No clear effect |
| add audio to strongest baseline GLC by simple concatenation (ablation) | GLC · Ego4D | F1 score: 37.8 -> 38.1 | No clear effect |
| CSTS versus GLC | CSTS · Ego4D | F1 score: 37.8 -> 39.7 | Improved |
| CSTS versus DFG+ (prior anticipation SOTA) | CSTS · Aria | F1 score: 57.6 -> 59.9 | Improved |
Closest papers by meaning
- Gaze Beyond the Frame: Forecasting Egocentric 3D Visual Span (NeurIPS 2025 · similarity 0.90)
- EyEar: Learning Audio Synchronized Human Gaze Trajectory Based on Physics-Informed Dynamics (AAAI 2025 · similarity 0.88)
- EgoSound: Benchmarking Sound Understanding in Egocentric Videos (CVPR 2026 · similarity 0.88)
- Gaze-VLM: Bridging Gaze and VLMs through Attention Regularization for Egocentric Understanding (NeurIPS 2025 · similarity 0.88)
- Sound Bridge: Associating Egocentric and Exocentric Videos via Audio Cues (CVPR 2025 · similarity 0.87)
- GazeFlow: From Human Gaze Behavior to Generative Egocentric Gaze Prediction (NeurIPS 2026 · similarity 0.87)
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.
