List your company
ResearchFormerly WizWand

Image Classification on Places 365

65.1Top-1 AccOmniVec2
37.394444.587251.7858.9728Nov 11, 2021Aug 11, 2022May 12, 2023Feb 10, 2024Nov 10, 2024Aug 11, 2025May 12, 2026
Updated 6mo ago

Evaluation results

MethodLinks
OmniVec22025.0765.1——
OmniVec2025.0763.5——
InternImage2025.0761.2——
MetaFormer2025.0760.7——
MAEBackbone=ViT-H, Input Resolution=448, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.1160.3——
Omnivore2025.0759.9——
MAEBackbone=ViT-H, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.1159.8——
Omni-MAE2025.0759.4——
MAEBackbone=ViT-L, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.1159.4——
EfficientNetBackbone=B82025.0758.6——
Previous bestPre-train Data=3.5 billion images2021.1158——
MAEBackbone=ViT-B, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.1157.9——
SigLIP 2-B/16Evaluation Protocol=Linear Probe2026.0556.18——
VECA-B/16Evaluation Protocol=Linear Probe, Core tokens (C)=642026.0555.82——
DFNCLIP-B/16Evaluation Protocol=Linear Probe2026.0555.79——
DINOv3-B/16Evaluation Protocol=Linear Probe2026.0555.39——
OpenCLIP-B/16Evaluation Protocol=Linear Probe2026.0555.35——
VECA-B/16Evaluation Protocol=Linear Probe, Core tokens (C)=82026.0555.3——
CLIP-B/16Evaluation Protocol=Linear Probe2026.0555.27——
DINOv2-reg-B/14Evaluation Protocol=Linear Probe2026.0554.81——
AM-RADIOv2.5-B/16Evaluation Protocol=Linear Probe2026.0554.13——
DINOv2-B/14Evaluation Protocol=Linear Probe2026.0553.37——
SALF-CBMSparse final layer=false2025.0249.38——
StandardSparse final layer=false2025.0248.56——
SALF-CBMSparse final layer=true2025.0246.73——
LF-CBMSparse final layer=true2025.0243.68——
StandardSparse final layer=true2025.0238.46——
Individual NetworksBackbone=VGG-16, # Models=2, Size=1,074 MB2018.01—46.3516.14
Jointly Trained NetworkBackbone=VGG-16, # Models=1, Size=537 MB2018.01—45.9815.59
PackNetBackbone=VGG-16, # Models=1, Size=554 MB2018.01—46.6415.92
PiggybackBackbone=VGG-16, # Models=1, Size=554 MB2018.01—46.7116.18