域适应
域适应(Domain Adaptation)是指在不同数据分布之间调整模型的任务。其核心目标是使机器学习模型能够泛化到目标域,并有效处理源域与目标域之间的分布差异,从而提升模型在新环境下的性能和鲁棒性。该技术在跨域数据应用中具有重要价值,可广泛应用于图像识别、自然语言处理等领域。
Office-31
PMTrans
SYNTHIA-to-Cityscapes
HALO
Office-Home
SWG
VisDA2017
DePT
GTA5 to Cityscapes
ProDA
ImageCLEF-DA
SPL
Cityscapes to ACDC
Refign (DAFormer)
MNIST-to-USPS
DFA-MCD
SVHN-to-MNIST
Mean teacher
USPS-to-MNIST
SVNH-to-MNIST
SRDA (RAN)
MoLane
MuLane
UFLD-SGADA-ResNet32
Office-Caltech
SPL
TuLane
Cityscapes-to-FoggyZurich
BWG
GTAV+Synscapes to Cityscapes
DDB
SYNSIG-to-GTSRB
DFA-MCD
Cityscapes-to-FoggyDriving
BWG
GTA5+Synscapes to Cityscapes
MRNet
HMDBfull-to-UCF
MNIST-to-MNIST-M
DRANet
Panoptic SYNTHIA-to-Cityscapes
Panoptic SYNTHIA-to-Mapillary
MC-PanDA
UCF --> HMDB (full)
UNITE
UCF-to-HMDBfull
GTAV to Cityscapes+Mapillary
Rein
HMDB --> UCF (full)
TA3N
Synth Digits-to-SVHN
DSN (DANN)
Synth Signs-to-GTSRB
Mean teacher
DomainNet
SFDA2
HMDBsmall-to-UCF
Olympic-to-HMDBsmall
Synscapes-to-Cityscapes
UCF-to-HMDBsmall
UCF-to-Olympic
TemPooling + RevGrad
Rotating MNIST
PCIDA
Canon RAW Low Light
Comic2k
Foggy Cityscapes
GTA-to-FoggyCityscapes
LeukemiaAttri
ConfMix [23] L_100x_C2
MNIST-M-to-MNIST
MSDA
Nikon RAW Low Light
Noisy-Amazon (20%)
Noisy-Amazon (45%)
Noisy-MNIST-to-SYND
Noisy-SYND-to-MNIST
Office-Caltech-10
MEDA
PACS
SSGEN
S2RDA-49
S2RDA-MS-39
PGA
Sim10k
Synth Objects-to-LINEMOD
DSN (DANN)
SYNTHIA-to-FoggyCityscapes
SYNTHIA-to-Cityscapes Labels
MRNet
VIPER-to-Cityscapes