Emotion Classification on FI
88.64AccuracyCLIP
Evaluation results
| Method | Links | |
|---|---|---|
| CLIPTraining Dataset=Selecta, Architecture=CLIP+MLP2025.12 | 88.64 | |
| CLIPTraining Dataset=EmoSet, Architecture=CLIP+MLP2025.12 | 86.3 | |
| SwinBTraining Dataset=Selecta, Architecture=Swin-B2025.12 | 72.64 | |
| ResNet-101Training Dataset=Selecta, Architecture=ResNet-1012025.12 | 71.79 | |
| SwinBTraining Dataset=EmoSet, Architecture=Swin-B2025.12 | 69.91 | |
| ResNet-101Training Dataset=EmoSet, Architecture=ResNet-1012025.12 | 67.91 | |
| ViT-B/16Enhancement strategy=w/ NS-LF2026.04 | 67.4 | |
| ViT-B/16Enhancement strategy=NS-LF2026.04 | 67.4 | |
| ResNet-50Enhancement strategy=w/ NS-LF2026.04 | 67.2 | |
| ResNet-50Enhancement strategy=NS-LF2026.04 | 67.2 | |
| Swin-TEnhancement strategy=w/ NS-LF2026.04 | 67 | |
| Swin-TEnhancement strategy=NS-LF2026.04 | 67 | |
| ResNet-50Enhancement strategy=Basic2026.04 | 65.8 | |
| ResNet-50Enhancement strategy=Basic2026.04 | 65.8 | |
| ViT-B/16Enhancement strategy=Basic2026.04 | 65.6 | |
| ViT-B/16Enhancement strategy=Basic2026.04 | 65.6 | |
| DenseNet-121Enhancement strategy=w/ NS-LF2026.04 | 64.4 | |
| Swin-TEnhancement strategy=Basic2026.04 | 64.4 | |
| DenseNet-121Enhancement strategy=NS-LF2026.04 | 64.4 | |
| Swin-TEnhancement strategy=Basic2026.04 | 64.4 | |
| Avg.Enhancement strategy=w/ NS-LF2026.04 | 64.2 | |
| DenseNet-121Enhancement strategy=Basic2026.04 | 63.2 | |
| DenseNet-121Enhancement strategy=Basic2026.04 | 63.2 | |
| Avg.Enhancement strategy=Basic2026.04 | 62.5 | |
| VGG-16Enhancement strategy=w/ NS-LF2026.04 | 61.4 | |
| VGG-16Enhancement strategy=NS-LF2026.04 | 61.4 | |
| VGG-16Enhancement strategy=Basic2026.04 | 59.7 | |
| VGG-16Enhancement strategy=Basic2026.04 | 59.7 | |
| AlexNetEnhancement strategy=w/ NS-LF2026.04 | 57.9 | |
| AlexNetEnhancement strategy=NS-LF2026.04 | 57.9 | |
| AlexNetEnhancement strategy=Basic2026.04 | 56.4 | |
| AlexNetEnhancement strategy=Basic2026.04 | 56.4 |