Image Classification on WikiArt GAN, WikiArt Style, Flowers, and PCam Average
803Throughput (Img/s)ResNet-18 + SpiralFovea+Mamba
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| ResNet-18 + SpiralFovea+MambaBackbone=ResNet-18, Tokenisation=SpiralFovea+Mamba, Tokens=≤78, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 803 | — | 86.8 | |
| ResNet-18 + MLP (uniform)Backbone=ResNet-18, Tokenisation=MLP (uniform), GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 724 | — | 83.8 | |
| DINO-ViT-S + SpiralFoveaBackbone=DINO-ViT-S, Tokenisation=SpiralFovea, Tokens=≤78, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 721 | 0.8 | 89.1 | |
| ViT-S/16 + SpiralFoveaBackbone=ViT-S/16, Tokenisation=SpiralFovea, Tokens=≤78, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 714 | 0.8 | 87.6 | |
| ResNet-18 + Mamba (uniform)Backbone=ResNet-18, Tokenisation=Mamba (uniform), GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 681 | — | 85.1 | |
| DINO-ViT-S + UniformBackbone=DINO-ViT-S, Tokenisation=Uniform, Tokens=196, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 612 | 4.6 | 87.4 | |
| ViT-S/16 + UniformBackbone=ViT-S/16, Tokenisation=Uniform, Tokens=196, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 598 | 4.6 | 85.5 | |
| ViT-B/16 + SpiralFoveaBackbone=ViT-B/16, Tokenisation=SpiralFovea, Tokens=≤78, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 531 | 2.82 | 88.2 | |
| ViT-B/16 + UniformBackbone=ViT-B/16, Tokenisation=Uniform, Tokens=196, GPU Hardware=Tesla T4, Batch Size=32, Input Resolution=224x2242026.07 | 412 | 17.5 | 86.4 |