Multimodal Depression Detection on PDCD 2025 (val)
81.25AccuracyACMG w Cross-modal (Qwen)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ACMG w Cross-modal (Qwen)Pre-trained Language Model=Qwen-embedding-0.6B, Gating configuration=Cross-modal2026.04 | 81.25 | 80.77 | 80.61 | 80.93 | |
| ACMG w Unimodal (Qwen)Pre-trained Language Model=Qwen-embedding-0.6B, Gating configuration=Unimodal2026.04 | 80.88 | 80.58 | 80.31 | 80.85 | |
| Transformer(Qwen)Pre-trained Language Model=Qwen-embedding-0.6B2026.04 | 79.78 | 79.48 | 79.37 | 79.59 | |
| ACMG w Cross-modal (RoBERTa)Pre-trained Language Model=RoBERTa, Gating configuration=Cross-modal2026.04 | 78.68 | 78.33 | 78.33 | 78.33 | |
| ACMG w Unimodal (RoBERTa)Pre-trained Language Model=RoBERTa, Gating configuration=Unimodal2026.04 | 77.57 | 77.33 | 77.07 | 77.59 | |
| Transformer(RoBERTa)Pre-trained Language Model=RoBERTa2026.04 | 75.74 | 76.17 | 75.64 | 76.7 | |
| Gomez-Zaragoza et al. (2025)2026.04 | 75.37 | 75.55 | 76.03 | 75.07 | |
| Al Hanai et al. (2018)2026.04 | 72.43 | 72.46 | 72.25 | 72.67 |