图像聚类
图像聚类是计算机视觉领域的一项重要任务,旨在将图像数据集划分为语义上有意义的簇,而无需访问真实标签。该任务通过无监督学习方法,自动发现图像中的内在结构和模式,从而实现对未标注图像的有效组织和管理。图像聚类在图像检索、数据挖掘和内容分析等应用场景中具有重要价值。
CIFAR-10
SCAN
CIFAR-100
HUME
STL-10
RUC
Imagenet-dog-15
MAE-CT (best)
ImageNet-10
DCCM
MNIST-full
SPC
USPS
SPC
Tiny-ImageNet
PRO-DSC
Fashion-MNIST
N2D (UMAP)
ImageNet
TURTLE (CLIP + DINOv2)
MNIST-test
DynAE
coil-100
JULE-RC
Extended Yale-B
DMSC
Coil-20
JULE-RC
ImageNet-100
ImageNet-200
TEMI CLIP ViT-L (openai)
ImageNet-50
Stanford Cars
FineGAN
CMU-PIE
CUB Birds
FineGAN
Stanford Dogs
FineGAN
UMist
J-DSSC (Scattered)
YouTube Faces DB
JULE-RC
coil-40
A-DSSC (Scattered)
FRGC
DEPICT
HAR
N2D (UMAP)
MNIST
DTD
TURTLE (CLIP + DINOv2)
EMNIST-Balanced
AE+SNNL
LetterA-J
DDC-DA
pendigits
N2D (UMAP)
UCF101
ARL Polarimetric Thermal Face Dataset
Birdsnap
Caltech-101
CARS196
CIFAR-20
CLEVR Counts
Country211
CUB-200-2011
EuroSAT
FER2013
FGVC Aircraft
Flowers-102
Food-101
GTSRB
Hateful Memes
imagenet-1k
TAC
Kinetics-700
KITTI
Oxford-IIIT Pets
PCam
Rendered SST2
TURTLE (CLIP + DINOv2)
RESISC45
SUN397