Fine-Grained Image Classification on FGVC 1.0 (test)
89.9CUB-2011GPS
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| GPSBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.772023.12 | 89.9 | 86.7 | 99.7 | 92.2 | 90.4 | 91.78 | |
| SSFBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.452023.12 | 89.5 | 85.7 | 99.6 | 89.6 | 89.2 | 90.72 | |
| SPT-AdapterBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.472023.12 | 89.1 | 83.3 | 99.2 | 91.1 | 86.2 | 89.78 | |
| SPT-LoRABackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.62023.12 | 88.6 | 83.4 | 99.5 | 91.4 | 87.3 | 90.04 | |
| VPT-DeepBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.992023.12 | 88.5 | 84.2 | 99 | 90.2 | 83.6 | 89.11 | |
| BiasBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.332023.12 | 88.4 | 84.2 | 98.8 | 91.2 | 79.4 | 88.4 | |
| FullBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=1002023.12 | 87.3 | 82.7 | 98.8 | 89.4 | 84.5 | 88.54 | |
| AdapterBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.482023.12 | 87.1 | 84.3 | 98.5 | 89.8 | 68.6 | 85.66 | |
| VPT-ShallowBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.292023.12 | 86.7 | 78.8 | 98.4 | 90.7 | 68.7 | 84.62 | |
| LoRABackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.92023.12 | 85.6 | 79.8 | 98.9 | 87.6 | 72 | 84.78 | |
| LinearBackbone=ViT-B/16, Pre-trained=ImageNet-21K, Params (%)=0.212023.12 | 85.3 | 75.9 | 97.9 | 86.2 | 51.3 | 79.32 |