Research map / Video / Action recognition
An End-to-End Visual-Audio Attention Network for Emotion Recognition in User-Generated Videos
Sicheng Zhao, Yunsheng Ma, Yang Gu, Jufeng Yang, Tengfei Xing, Pengfei Xu, Runbo Hu, Hua Chai, Kurt Keutzer
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 |
|---|---|---|---|
| use VAANet instead of the strongest baseline | Video emotion recognition · VideoEmotion-8 | accuracy: 52.5 -> 54.5 | Improved |
| use VAANet instead of the strongest baseline | Video emotion recognition · Ekman-6 | accuracy: 54.4 -> 55.3 | Improved |
| add visual channel-wise attention to visual spatial attention (ablation) | VAANet · VideoEmotion-8 | average accuracy: 51.7 -> 52.6 | Improved |
| add visual temporal attention to visual spatial and channel-wise attention (ablation) | VAANet · VideoEmotion-8 | average accuracy: 52.6 -> 53.6 | Improved |
| add audio temporal attention to visual attentions (ablation) | VAANet · VideoEmotion-8 | average accuracy: 53.6 -> 54.5 | Improved |
| add visual channel-wise attention to visual spatial attention (ablation) | VAANet · Ekman-6 | average accuracy: 50.8 -> 53.4 | Improved |
| add visual temporal attention to visual spatial and channel-wise attention (ablation) | VAANet · Ekman-6 | average accuracy: 53.4 -> 54.5 | Improved |
| add audio temporal attention to visual attentions (ablation) | VAANet · Ekman-6 | average accuracy: 54.5 -> 55.3 | Improved |
| replace CE with PCCE using visual attentions only (ablation) | VAANet · VideoEmotion-8 | average accuracy: 51.9 -> 53.6 | Improved |
| replace CE with PCCE using visual and audio attentions (ablation) | VAANet · VideoEmotion-8 | average accuracy: 53.9 -> 54.5 | Improved |
| replace CE with PCCE using visual attentions only (ablation) | VAANet · Ekman-6 | average accuracy: 52.0 -> 54.5 | Improved |
| replace CE with PCCE using visual and audio attentions (ablation) | VAANet · Ekman-6 | average accuracy: 54.6 -> 55.3 | Improved |
Closest papers by meaning
- MDAN: Multi-Level Dependent Attention Network for Visual Emotion Analysis (CVPR 2022 · similarity 0.88)
- Building a Large Scale Dataset for Image Emotion Recognition: The Fine Print and The Benchmark (AAAI 2016 · similarity 0.86)
- Weakly Supervised Video Emotion Detection and Prediction via Cross-Modal Temporal Erasing Network (CVPR 2023 · similarity 0.86)
- Learning Visual Emotion Representations From Web Data (CVPR 2020 · similarity 0.86)
- MIMAMO Net: Integrating Micro- and Macro-Motion for Video Emotion Recognition (AAAI 2020 · similarity 0.86)
- EmoVIT: Revolutionizing Emotion Insights with Visual Instruction Tuning (CVPR 2024 · 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.
