Long Tail Learning On Cifar 100 Lt R 50

Metriken

Error Rate

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	Error Rate	Paper Title	Repository
DeiT-LT	39.5	DeiT-LT Distillation Strikes Back for Vision Transformer Training on Long-Tailed Datasets	-
BCL(ResNet-32)	43.4	Balanced Contrastive Learning for Long-Tailed Visual Recognition	-
ConCutMix	-	Enhanced Long-Tailed Recognition with Contrastive CutMix Augmentation	-
MetaSAug-LDAM	47.73	MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition	-
LDAM-DRW + SSP	52.89	Rethinking the Value of Labels for Improving Class-Imbalanced Learning	-
LIFT (ViT-B/16, CLIP)	16.9	Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts	-
GCL	46.4	Long-tailed Visual Recognition via Gaussian Clouded Logit Adjustment
PC	42.25	Learning Prototype Classifiers for Long-Tailed Recognition	-
Difficulty-Net	43.1	Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition	-
LDAM-DRW-RSG	51.5	RSG: A Simple but Effective Module for Learning Imbalanced Datasets	-
GLMC (ResNet-34, channel x4)	36.15	Global and Local Mixture Consistency Cumulative Learning for Long-tailed Visual Recognitions	-
LIFT (ViT-B/16, ImageNet-21K pre-training)	9.8	Long-Tail Learning with Foundation Model: Heavy Fine-Tuning Hurts	-
GLMC + SAM	34.72	Escaping Saddle Points for Effective Generalization on Class-Imbalanced Data	-
OPeN (WideResNet-28-10)	40.2	Pure Noise to the Rescue of Insufficient Data: Improving Imbalanced Classification by Training on Random Noise Images	-
MiSLAS	47.7	Improving Calibration for Long-Tailed Recognition	-
NCL(ResNet32)	43.2	Nested Collaborative Learning for Long-Tailed Visual Recognition	-
TADE	46.1	Self-Supervised Aggregation of Diverse Experts for Test-Agnostic Long-Tailed Recognition	-
Hybrid-PSC	51.07	Contrastive Learning based Hybrid Networks for Long-Tailed Image Classification	-
CBD+TailCalibX	49.1	Feature Generation for Long-tail Classification	-
GML (ResNet-32)	41.9	Long-Tailed Recognition by Mutual Information Maximization between Latent Features and Ground-Truth Labels	-

0 of 25 row(s) selected.