List your company
ResearchFormerly WizWand

Image Recognition on ImageNet-1K

91Top-1 AccuracyAbsolute SOTA
62.576869.955977.33584.7141Mar 19, 2024Jul 18, 2024Nov 17, 2024Mar 18, 2025Jul 18, 2025Nov 16, 2025Mar 18, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Absolute SOTA2024.0691
MiCoBackbone=ViT-g2024.0689.8
VT-FSLLearning Paradigm=Training-based2026.0389.75
AIMv2architecture=ViT-3B/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=448px, Pre-training samples=12B2024.1189.5
AIMv2architecture=ViT-3B/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px, Pre-training samples=12B2024.1188.5
Meta-TransformerBackbone=ViT-L2024.0688.1
AIMv2architecture=ViT-1B/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px, Pre-training samples=12B2024.1188.1
AIMv2architecture=ViT-H/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px, Pre-training samples=12B2024.1187.5
SigLIParchitecture=ViT-So400m/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1187.3
DINOv2architecture=ViT-g/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1187.2
Ours w/ PointBindSetting=Fine-tuning2024.0386.94
DFN-CLIParchitecture=ViT-H/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1186.9
AIMv2architecture=ViT-L/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px, Pre-training samples=12B2024.1186.6
DFN-CLIParchitecture=ViT-L/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1186.5
SigLIParchitecture=ViT-L/16, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1186.5
OAI CLIParchitecture=ViT-L/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1185.7
SARELearning Paradigm=Training-based2026.0385.06
AIMv1architecture=ViT-7B/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1184
Ours w/ PointBindSetting=Zero-shot2024.0383.25
Meta-TransformerSetting=Fine-tuning2024.0383.1
MAEarchitecture=ViT-2B/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1182.2
UniFGVCLearning Paradigm=Training-free, Method Strategy=Reasoning-based2026.0381.1
ImageBindBackbone=ViT-H2024.0680.2
ImageBindSetting=Fine-tuning, Protocol=linear2024.0380.19
PointBindSetting=Fine-tuning, Protocol=linear2024.0380.19
PointBind (+Event)Setting=Fine-tuning2024.0380.19
AIMv1architecture=ViT-H/14, Evaluation protocol=Attentive probing, Backbone status=Frozen, Resolution=224px2024.1178.5
ImageBindSetting=Zero-shot2024.0377.7
PointBindSetting=Zero-shot2024.0377.7
PointBind (+Event)Setting=Zero-shot2024.0377.7
CascadeVLMLearning Paradigm=Training-free, Method Strategy=Reasoning-based2026.0373.98
ProtoMMLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0372.76
SCANLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0372.14
AutoSEPLearning Paradigm=Training-free, Method Strategy=Reasoning-based2026.0371.45
FineDeficsLearning Paradigm=Training-based2026.0371.41
AWTLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0371.32
RARLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0370.26
E-FineRLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0369.02
FineRLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0368.65
Qwen2.5-VL-7BLearning Paradigm=Backbone2026.0368.4
SuS-X-LCLearning Paradigm=Training-free, Method Strategy=Retrieval-based2026.0366.87
CLIP-B/32Learning Paradigm=Backbone2026.0363.67