List your company
ResearchFormerly WizWand

Image Classification on FGVC (test)

89.11AccuracyVPT-DEEP
14.573233.924153.27572.6259Mar 23, 2022Nov 15, 2022Jul 10, 2023Mar 3, 2024Oct 26, 2024Jun 20, 2025Feb 12, 2026
Updated 8mo ago

Evaluation results

MethodLinks
VPT-DEEPTotal parameters=1.18x, Scope: Input=true, Scope: Backbone=true, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0389.114
FULLTotal parameters=24.02x, Scope: Input=false, Scope: Backbone=false, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0388.54—
BIASTotal parameters=1.05x, Scope: Input=false, Scope: Backbone=true, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0388.413
ADAPTERTotal parameters=1.23x, Scope: Input=false, Scope: Backbone=true, Extra parameters=true, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0385.662
VPT-SHALLOWTotal parameters=1.04x, Scope: Input=true, Scope: Backbone=false, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0384.621
PARTIAL-1Total parameters=3.00x, Scope: Input=false, Scope: Backbone=false, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0382.630
LoRA_mul + VPT_add + PACEShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0981.9—
LoRA_mul + VPT_addShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0981.2—
LoRA_add + PACEShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0980.8—
MLP-3Total parameters=1.35x, Scope: Input=false, Scope: Backbone=false, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0379.80
LINEARTotal parameters=1.02x, Scope: Input=false, Scope: Backbone=false, Extra parameters=false, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0379.320
LoRA_addShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0978.9—
SIDETUNETotal parameters=3.69x, Scope: Input=false, Scope: Backbone=true, Extra parameters=true, Backbone: ViT-B/16=85.8M, Pre-training: supervised ImageNet-21k=true2022.0378.350
VPT_add + PACEShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0977.2—
VPT_addShot=16, Backbone=ViT-B/16, Pre-trained=ImageNet-21K2024.0976.7—
FullBackbone=ViT-Large, # Params=303M2024.0767.38—
C3ABackbone=ViT-Large, Config=b=1024/16, # Params=0.79M2024.0763.8—
LoRABackbone=ViT-Large, Rank=r=16, # Params=1.57M2024.0763.64—
FullBackbone=ViT-Base, # Params=85.8M2024.0761.6—
LoRABackbone=ViT-Base, Rank=r=16, # Params=0.59M2024.0756.64—
C3ABackbone=ViT-Base, Config=b=768/12, # Params=0.22M2024.0754.31—
ARFCBackbone=SigLip 2 (ViT-B/16), Compression Ratio=50%, Evaluation Protocol=Zero-shot2026.0229—
ARFCBackbone=CN-CLIP (ViT-H/14), Compression Ratio=50%, Evaluation Protocol=Zero-shot2026.0226.4—
HeadBackbone=ViT-Large2024.0724.62—
HeadBackbone=ViT-Base2024.0717.44—