Models Generic DINOv2 Large

Meta

DINOv2 Large

Self-supervised vision transformer from Meta AI.

Meta Open Source Vision Only

HuggingFace

HuggingFace GitHub

GitHub

arXiv

arXiv

Generic

Model Type

#9

Overall Rank

36.93%

Avg. R@1

23.80%

Avg.
mAP@20

1024

Embedding
Size

224

Input
Size

7

Datasets

Performance Across Datasets

Dataset	Category	R@1	R@5	mAP
TOPEX	Industrial	60.69%	77.20%	34.62%
Stanford Online Products	E-commerce	56.34%	67.74%	31.91%
Products-10K	E-commerce	50.19%	66.38%	29.02%
Furniture v1	Retail	40.16%	59.84%	24.62%
DIY v1	Hardware/DIY	18.17%	35.12%	24.95%
Automotive v1	Automotive	19.30%	31.81%	18.76%
Clips-and-Connectors v1	Industrial	13.66%	26.43%	2.71%
Average		36.93%	52.07%	23.80%

All Models View Full Results