Research map / Probabilistic and scientific modeling / Causal discovery
Context De-Confounded Emotion Recognition
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 CCIM to EmotiCon | EmotiCon · EMOTIC | mAP: 35.28 -> 39.13 | Helped |
| Add CCIM to EmotiCon | EmotiCon · GroupWalk | mAP: 65.58 -> 69.31 | Helped |
| Add CCIM to EmotiCon | EmotiCon · CAER-S | accuracy: 88.65 -> 91.17 | Helped |
| Use a learned context-prototype dictionary rather than a random dictionary (ablation) | CCIM with EMOT-Net and EmotiCon · EMOTIC, CAER-S, GroupWalk | mAP / accuracy: EMOT-Net: 26.56 vs 30.88, 73.36 vs 75.82, 57.45 vs 60.74; EmotiCon: 35.12 vs 39.13, 87.34 vs 91.17, 65.62 vs 69.31 | Helped |
| Use Places365 rather than ImageNet pre-training for context features (ablation) | CCIM with EMOT-Net and EmotiCon · EMOTIC, CAER-S, GroupWalk | mAP / accuracy: EMOT-Net: 28.72 vs 30.88, 74.75 vs 75.82, 58.96 vs 60.74; EmotiCon: 37.48 vs 39.13, 90.46 vs 91.17, 68.28 vs 69.31 | Helped |
| Replace dot-product attention with additive attention (ablation) | CCIM with EMOT-Net and EmotiCon · EMOTIC, CAER-S, GroupWalk | mAP / accuracy: EMOT-Net: 30.79 vs 30.88, 75.64 vs 75.82, 60.85 vs 60.74; EmotiCon: 39.16 vs 39.13, 91.08 vs 91.17, 69.26 vs 69.31 | No clear effect |
| Add attention weights λᵢ to confounder integration (ablation) | CCIM with EMOT-Net · EMOTIC | mAP: 30.05 -> 30.88 | Helped |
| Add prior probabilities P(zᵢ) to confounder integration (ablation) | CCIM with EmotiCon · CAER-S | accuracy: 90.06 -> 91.17 | Helped |
| Mask the target subject when constructing context prototypes (ablation) | CCIM with EmotiCon · GroupWalk | mAP: 67.79 -> 69.31 | Helped |
Closest papers by meaning
- Robust Emotion Recognition in Context Debiasing (CVPR 2024 · similarity 0.94)
- CocoER: Aligning Multi-Level Feature by Competition and Coordination for Emotion Recognition (CVPR 2025 · similarity 0.85)
- Facial-R1: Aligning Reasoning and Recognition for Facial Emotion Analysis (AAAI 2026 · similarity 0.85)
- Bridge Then Begin Anew: Generating Target-Relevant Intermediate Model for Source-Free Visual Emotion Adaptation (AAAI 2025 · similarity 0.85)
- AffectGPT: A New Dataset, Model, and Benchmark for Emotion Understanding with Multimodal Large Language Models (ICML 2025 · similarity 0.84)
- VidEmo: Affective-Tree Reasoning for Emotion-Centric Video Foundation Models (NeurIPS 2025 · similarity 0.84)
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.