Facial Expression Classification on Independent seven-class facial-expression benchmark (test)
6,394Correct CountFine-tuned model
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Fine-tuned modelFine-tuning protocol=Fine-tuned on target training split, Backbone=Hugging Face ViT checkpoint, Batch size=32, Optimizer=AdamW, Learning rate=5 × 10−5, Input resolution=224 × 224 × 32026.05 | 6,394 | 6,767 | 94.49 | |
| Baseline modelFine-tuning protocol=Non-fine-tuned, Backbone=Hugging Face ViT checkpoint2026.05 | 4,037 | 6,767 | 59.66 |