Conversation Detection on Smartwatch Audio and Motion Sensing L-LOGO 1.0
82Macro-F1Pure-Acoustic Model + CNN+Attention
Evaluation results
| Method | Links | |
|---|---|---|
| Pure-Acoustic Model + CNN+AttentionModality=Audio+Motion, Fusion=Concatenation2025.07 | 82 | |
| Pure-Acoustic Model + SCNNBModality=Audio+Motion, Fusion=Concatenation2025.07 | 77.5 | |
| Pure-Acoustic ModelModality=Audio, Fusion=N/A2025.07 | 74.4 | |
| Pure-Acoustic Model + SCNNBModality=Audio+Motion, Fusion=Softmax Averaging2025.07 | 67.9 | |
| Pure-Acoustic Model + SCNNBModality=Audio+Motion, Fusion=Self-Attention2025.07 | 62.9 | |
| Pure-Acoustic Model + CNN+AttentionModality=Audio+Motion, Fusion=Softmax Averaging2025.07 | 62.4 | |
| Pure-Acoustic Model + CNN+AttentionModality=Audio+Motion, Fusion=Self-Attention2025.07 | 59.6 | |
| SCNNBModality=Motion, Fusion=N/A2025.07 | 51.6 | |
| CNN+AttentionModality=Motion, Fusion=N/A2025.07 | 49.6 |