Long Tail Learning On Imagenet Lt

评估指标

Top-1 Accuracy

评测结果

各个模型在此基准测试上的表现结果

		Paper Title	Repository
LIFT (ViT-L/14)	82.9	Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts
µ2Net+ (ViT-L/16)	82.5	A Continual Development Methodology for Large-scale Multitask Dynamic ML Systems
MAM (ViT-B/16)	82.3	Improving Image Recognition by Retrieving from Web-Scale Image-Text Data	-
LIFT (ViT-B/16)	78.3	Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts
VL-LTR (ViT-B-16)	77.2	VL-LTR: Learning Class-wise Visual-Linguistic Representation for Long-Tailed Visual Recognition
BALLAD(ResNet-50×16)	76.5	A Simple Long-Tailed Recognition Baseline via Vision-Language Model
BALLAD(ViT-B-16)	75.7	A Simple Long-Tailed Recognition Baseline via Vision-Language Model
BALLAD(ResNet-101)	70.5	A Simple Long-Tailed Recognition Baseline via Vision-Language Model
VL-LTR (ResNet-50)	70.1	VL-LTR: Learning Class-wise Visual-Linguistic Representation for Long-Tailed Visual Recognition
BALLAD(ResNet-50)	67.2	A Simple Long-Tailed Recognition Baseline via Vision-Language Model
GPaCo (2-ResNeXt101-32x4d)	63.2	Generalized Parametric Contrastive Learning
MDCS (ResNeXt-50)	61.8	MDCS: More Diverse Experts with Consistency Self-distillation for Long-tailed Recognition
TADE(ResNeXt101-32x4d)	61.4	Self-Supervised Aggregation of Diverse Experts for Test-Agnostic Long-Tailed Recognition
ProCo (ResNet50)	60.2	Probabilistic Contrastive Learning for Long-Tailed Visual Recognition
PaCo(ResNeXt101-32x4d)	60.0	Parametric Contrastive Learning
DeiT-LT	59.1	DeiT-LT Distillation Strikes Back for Vision Transformer Training on Long-Tailed Datasets
APA (SE-ResNext-50)	59.1	Adaptive Parametric Activation
TADE(ResNeXt-50)	58.8	Self-Supervised Aggregation of Diverse Experts for Test-Agnostic Long-Tailed Recognition
GML (ResNeXt-50)	58.8	Long-Tailed Recognition by Mutual Information Maximization between Latent Features and Ground-Truth Labels
DirMixE(ResNeXt-50)	58.61	Harnessing Hierarchical Label Distribution Variations in Test Agnostic Long-tail Recognition

0 of 69 row(s) selected.