List your company
ResearchFormerly WizWand

Image Classification on CUB200 (Accuracy, Complexity)

91.6AccuracyViT-NeT
28.440844.837961.23577.6321Aug 26, 2025Oct 13, 2025Dec 1, 2025Jan 19, 2026Mar 9, 2026Apr 27, 2026Jun 15, 2026
Updated 4mo ago

Evaluation results

MethodLinks
ViT-NeTViT=true, no retraining of encoder=true, no training of clf head=false, backbone=ViT with DinoV2 weights2025.1291.6——
CBM-SuiteBackbone=ResNet18, Selection=Best-performing vision backbone–VLM pair2026.0386.73——
Full FT.Rank=-, Backbone=ViT-Base2026.0686.35——
ST-ProtoPNetViT=true, no retraining of encoder=true, no training of clf head=false, backbone=ViT with DinoV2 weights2025.1286.1——
SDS-LoRARank=8, Backbone=ViT-Base2026.0686.07——
SDS-LoRARank=32, Backbone=ViT-Base2026.0685.99——
LoRARank=32, Backbone=ViT-Base2026.0685.75——
DoRARank=8, Backbone=ViT-Base2026.0685.74——
LoRA+Rank=32, Backbone=ViT-Base2026.0685.68——
rsLoRARank=8, Backbone=ViT-Base2026.0685.67——
DoRARank=32, Backbone=ViT-Base2026.0685.67——
LoRARank=8, Backbone=ViT-Base2026.0685.65——
ProtoPoolViT=false, no retraining of encoder=false, no training of clf head=false2025.1285.5202—
rsLoRARank=32, Backbone=ViT-Base2026.0685.42——
LoRA+Rank=8, Backbone=ViT-Base2026.0685.28——
ProtoS-ViTViT=true, no retraining of encoder=true, no training of clf head=false, backbone=ViT with DinoV2 weights2025.1285.26—
PiSSARank=8, Backbone=ViT-Base2026.0685.17——
KMExViT=true, no retraining of encoder=true, no training of clf head=true, backbone=ViT with DinoV2 weights2025.12851—
PiSSARank=32, Backbone=ViT-Base2026.0684.9——
PIP-NetViT=false, no retraining of encoder=false, no training of clf head=false2025.1284.34—
ProtoTreeViT=false, no retraining of encoder=false, no training of clf head=false2025.1282.28.3—
KCCsViT=true, no retraining of encoder=true, no training of clf head=true, backbone=ViT with DinoV2 weights2025.1282.22.2—
LaBoSelection=Best-performing vision backbone–VLM pair2026.0381——
ProtoPNetViT=false, no retraining of encoder=false, no training of clf head=false2025.1279.22,000—
LFCBMSelection=Best-performing vision backbone–VLM pair2026.0374.31——
BPBackbone=MobileNetV2, # Layers=42025.0965.860.0157.42
BPBackbone=MobileNetV2, # Layers=22025.0964.370.0130.62
BPBackbone=ResNet34, # Layers=42025.0964.220.1249
BPBackbone=ResNet34, # Layers=22025.0964.10.0624.5
LANCEBackbone=ResNet34, # Layers=42025.0963.010.060.8
LANCEBackbone=ResNet34, # Layers=22025.0962.350.030.39
BPBackbone=ResNet18, # Layers=22025.0962.280.0624.5
HOSVDBackbone=ResNet34, # Layers=42025.0962.2812.250.84
LANCEBackbone=MobileNetV2, # Layers=42025.0961.210.010.15
HOSVDBackbone=ResNet34, # Layers=22025.0960.826.120.38
BPBackbone=ResNet18, # Layers=42025.0960.80.0961.25
HOSVDBackbone=MobileNetV2, # Layers=42025.0960.322.860.16
HOSVDBackbone=ResNet18, # Layers=22025.0960.026.120.73
LANCEBackbone=ResNet18, # Layers=22025.0959.870.030.73
LANCEBackbone=ResNet18, # Layers=42025.0959.50.051.36
HOSVDBackbone=ResNet18, # Layers=42025.0959.2615.561.41
AXISArchitecture=ViT-B-322025.0859.06——
ASIBackbone=ResNet34, # Layers=42025.0958.850.071.58
ASIBackbone=ResNet34, # Layers=22025.0958.770.030.6
LANCEBackbone=MobileNetV2, # Layers=22025.0958.180.010.14
HOSVDBackbone=MobileNetV2, # Layers=22025.0958.1111.870.13
ASIBackbone=ResNet18, # Layers=22025.0956.770.041.51
aTLASArchitecture=ViT-B-322025.0855.82——
ASIBackbone=ResNet18, # Layers=42025.0955.730.073.5
ASIBackbone=MobileNetV2, # Layers=42025.0950.690.060.63
ASIBackbone=MobileNetV2, # Layers=22025.0946.440.010.26
BPBackbone=MCUNet, # Layers=42025.0934.130.00117.57
BPBackbone=MCUNet, # Layers=22025.0933.05011.71
HOSVDBackbone=MCUNet, # Layers=42025.0931.722.590.14
LANCEBackbone=MCUNet, # Layers=22025.0931.4400.05
HOSVDBackbone=MCUNet, # Layers=22025.0931.221.930.06
LANCEBackbone=MCUNet, # Layers=42025.0930.8700.13