Emotion and Micro-expression Analysis on PRISM
80.2Macro-expression AccuracyMIND-LLM-8B
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| MIND-LLM-8BBackbone=Qwen3-8B2025.12 | 80.2 | 34.4 | 0.875 | 0.541 | |
| GPT-5Category=General LMMs2025.12 | 76.7 | 18.4 | 0.805 | 0.545 | |
| Emotion-LLaMA-7B-CI*Category=Emotion LMMs, Training Data=Fine-tuned on ConvoInsight-DB2025.12 | 73.5 | 15.2 | 0.781 | 0.483 | |
| Emotion-LLaMA-7BCategory=Emotion LMMs2025.12 | 64.9 | 1.3 | 0.538 | 0.268 | |
| MMAFFLM-7B-CI*Category=Emotion LMMs, Training Data=Fine-tuned on ConvoInsight-DB2025.12 | 64.8 | 12.4 | 0.759 | 0.399 | |
| MMAFFLM-7BCategory=Emotion LMMs2025.12 | 51.7 | 0.6 | 0.636 | 0.207 | |
| MSE-Adapter-7BCategory=Emotion LMMs2025.12 | 48.2 | 0.2 | 0.692 | 0.219 | |
| EmoViT-7BCategory=Emotion LMMs2025.12 | 47.1 | 0 | 0.415 | 0.305 | |
| LLaVA-v1.5-7BCategory=General LMMs2025.12 | 42.2 | 0 | 0.774 | 0.279 | |
| InstructBLIP-13BCategory=General LMMs2025.12 | 42 | 0 | 0.637 | 0.263 | |
| Qwen3-VL-8BCategory=General LMMs2025.12 | 41.6 | 0.9 | 0.708 | 0.25 | |
| Emotion-Qwen-7BCategory=Emotion LMMs2025.12 | 23.3 | 0 | 0.156 | 0.358 | |
| DeepSeek-VL-7BCategory=General LMMs2025.12 | 16.7 | 0 | 0.533 | 0.195 |