5-way 1-shot Few-Shot Classification on BSCD-FSL Suite 1.0 (test)
0.2365ChestX AccuracyLoRA-CLIP + Semantic Probe
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| LoRA-CLIP + Semantic Probebackbone=ViT/CLIP2026.05 | 0.2365 | 0.3877 | 0.8294 | 0.8511 | — | — | — | — | 0.5762 | |
| wave-SANBackbone=ResNet-10, Fine-tuning=false2023.02 | 0.2293 | 0.3335 | 0.6964 | 0.708 | 0.5025 | 0.3355 | 0.5775 | 0.4071 | 0.4737 | |
| AFABackbone=ResNet-10, Fine-tuning=false2023.02 | 0.2292 | 0.3321 | 0.6312 | 0.6761 | 0.4686 | 0.3425 | 0.5404 | 0.3676 | 0.4485 | |
| StyleAdvBackbone=ViT-small, Fine-tuning=false, Large-scale Pretraining=DINO/IN1K2023.02 | 0.2292 | 0.3305 | 0.7215 | 0.8122 | 0.8401 | 0.4048 | 0.7264 | 0.5552 | 0.5775 | |
| StyleAdv-FTBackbone=ViT-small, Fine-tuning=true, Large-scale Pretraining=DINO/IN1K2023.02 | 0.2292 | 0.3399 | 0.7493 | 0.8411 | 0.8401 | 0.4048 | 0.7264 | 0.5552 | 0.5857 | |
| StepSPTMark=TPAMI-25, backbone=ViT/CLIP2026.05 | 0.2284 | 0.3297 | 0.7001 | 0.8484 | — | — | — | — | 0.5266 | |
| CLIP-Adapter + Semantic Probebackbone=ViT/CLIP2026.05 | 0.2273 | 0.3283 | 0.7019 | 0.8203 | — | — | — | — | 0.5194 | |
| StyleAdvBackbone=ResNet-10, Fine-tuning=false2023.02 | 0.2264 | 0.3396 | 0.7094 | 0.7413 | 0.4849 | 0.3464 | 0.5858 | 0.4113 | 0.4806 | |
| StyleAdv-FTBackbone=ResNet-10, Fine-tuning=true2023.02 | 0.2264 | 0.3576 | 0.7292 | 0.8069 | 0.4849 | 0.3509 | 0.5858 | 0.4113 | 0.4941 | |
| LDCMark=CVPR-25, backbone=ViT/CLIP, Data augmentation=true2026.05 | 0.2232 | 0.3372 | 0.7439 | 0.8407 | — | — | — | — | 0.5362 | |
| Tip-AdapterMark=ECCV-22, backbone=ViT/CLIP, Data augmentation=true2026.05 | 0.2224 | 0.3268 | 0.7544 | 0.7715 | — | — | — | — | 0.5187 | |
| Tip-Adapter + Semantic Probebackbone=ViT/CLIP2026.05 | 0.2223 | 0.3188 | 0.7429 | 0.7819 | — | — | — | — | 0.5165 | |
| ATA-FTBackbone=ResNet-10, Fine-tuning=true2023.02 | 0.2215 | 0.3494 | 0.6862 | 0.7541 | 0.4623 | 0.3715 | 0.5418 | 0.3738 | 0.4701 | |
| LRPBackbone=ResNet-10, Fine-tuning=false2023.02 | 0.2211 | 0.3094 | 0.5499 | 0.5923 | 0.4829 | 0.3278 | 0.5483 | 0.3749 | 0.4258 | |
| ATABackbone=ResNet-10, Fine-tuning=false2023.02 | 0.221 | 0.3321 | 0.6135 | 0.6747 | 0.45 | 0.3361 | 0.5357 | 0.3442 | 0.4384 | |
| FWTBackbone=ResNet-10, Fine-tuning=false2023.02 | 0.2204 | 0.3158 | 0.6236 | 0.6636 | 0.4747 | 0.3161 | 0.5577 | 0.3595 | 0.4414 | |
| GNNBackbone=ResNet-10, Fine-tuning=false2023.02 | 0.22 | 0.3202 | 0.6369 | 0.6448 | 0.4569 | 0.3179 | 0.531 | 0.356 | 0.4355 | |
| Tip-AdapterMark=ECCV-22, backbone=ViT/CLIP2026.05 | 0.2196 | 0.3036 | 0.7205 | 0.7655 | — | — | — | — | 0.5023 | |
| LoRA-CLIPMark=CVPR-24, backbone=RN50/CLIP2026.05 | 0.2176 | 0.3201 | 0.5779 | 0.6524 | — | — | — | — | 0.442 | |
| PMFBackbone=ViT-small, Fine-tuning=true, Large-scale Pretraining=DINO/IN1K2023.02 | 0.2173 | 0.3036 | 0.7074 | 0.8079 | 0.7813 | 0.3724 | 0.7111 | 0.536 | 0.5546 | |
| LoRA-CLIPMark=CVPR-24, backbone=ViT/CLIP2026.05 | 0.2173 | 0.3445 | 0.8034 | 0.8495 | — | — | — | — | 0.5537 | |
| LP++Mark=CVPR-24, backbone=ViT/CLIP, Data augmentation=true2026.05 | 0.2172 | 0.3363 | 0.7305 | 0.8184 | — | — | — | — | 0.5256 | |
| AMU-TuningMark=CVPR-24, backbone=ViT/CLIP, Data augmentation=true2026.05 | 0.2156 | 0.3229 | 0.7224 | 0.802 | — | — | — | — | 0.5157 | |
| CLIP-AdapterMark=IJCV-24, backbone=ViT/CLIP2026.05 | 0.2151 | 0.3131 | 0.6783 | 0.8114 | — | — | — | — | 0.5044 | |
| Maple + Semantic Probebackbone=ViT/CLIP2026.05 | 0.2141 | 0.3427 | 0.8294 | 0.8283 | — | — | — | — | 0.5536 | |
| CoOp + Semantic Probebackbone=ViT/CLIP2026.05 | 0.2133 | 0.2964 | 0.6977 | 0.8093 | — | — | — | — | 0.5042 | |
| CoOpMark=IJCV-22, backbone=ViT/CLIP2026.05 | 0.2095 | 0.2947 | 0.6816 | 0.7927 | — | — | — | — | 0.4946 | |
| MapleMark=CVPR-23, backbone=ViT/CLIP2026.05 | 0.2089 | 0.311 | 0.7566 | 0.8187 | — | — | — | — | 0.5304 |