SOTAVerified

Long-tail Learning

Long-tailed learning, one of the most challenging problems in visual recognition, aims to train well-performing models from a large number of images that follow a long-tailed class distribution.

Papers

Showing 1–50 of 131 papers

TitleStatusHype
Visual Prompt TuningCode3
Probabilistic Contrastive Learning for Long-Tailed Visual RecognitionCode2
Adaptive Parametric ActivationCode2
BatchFormer: Learning to Explore Sample Relationships for Robust Representation LearningCode2
SURE: SUrvey REcipes for building reliable and robust deep networksCode2
Focal Loss for Dense Object DetectionCode2
Generalized Parametric Contrastive LearningCode2
Learning Transferable Visual Models From Natural Language SupervisionCode2
A Simple Episodic Linear Probe Improves Visual Recognition in the WildCode2
Long-tail learning via logit adjustmentCode1
Large-Scale Long-Tailed Recognition in an Open WorldCode1
Invariant Feature Learning for Generalized Long-Tailed ClassificationCode1
Decoupling Representation and Classifier for Long-Tailed RecognitionCode1
Learning From Multiple Experts: Self-paced Knowledge Distillation for Long-tailed ClassificationCode1
Learning Imbalanced Datasets with Label-Distribution-Aware Margin LossCode1
From Generalized zero-shot learning to long-tail with class descriptorsCode1
Improving Calibration for Long-Tailed RecognitionCode1
Improving Visual Prompt Tuning by Gaussian Neighborhood Minimization for Long-Tailed Visual RecognitionCode1
Global and Local Mixture Consistency Cumulative Learning for Long-tailed Visual RecognitionsCode1
Long-Tailed Recognition by Mutual Information Maximization between Latent Features and Ground-Truth LabelsCode1
Balanced Contrastive Learning for Long-Tailed Visual RecognitionCode1
Imagine by Reasoning: A Reasoning-Based Implicit Semantic Data Augmentation for Long-Tailed ClassificationCode1
CUDA: Curriculum of Data Augmentation for Long-Tailed RecognitionCode1
Predicting and Enhancing the Fairness of DNNs with the Curvature of Perceptual ManifoldsCode1
Balanced Product of Calibrated Experts for Long-Tailed RecognitionCode1
A Simple Long-Tailed Recognition Baseline via Vision-Language ModelCode1
Long-tailed Recognition by Routing Diverse Distribution-Aware ExpertsCode1
DeiT-LT Distillation Strikes Back for Vision Transformer Training on Long-Tailed DatasetsCode1
AUCSeg: AUC-oriented Pixel-level Long-tail Semantic SegmentationCode1
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed RecognitionCode1
Class-Balanced Loss Based on Effective Number of SamplesCode1
Distribution-Balanced Loss for Multi-Label Classification in Long-Tailed DatasetsCode1
Class-Balanced Distillation for Long-Tailed Visual RecognitionCode1
Long-Tailed Classification by Keeping the Good and Removing the Bad Momentum Causal EffectCode1
Learning Imbalanced Data with Vision TransformersCode1
Do Deep Networks Transfer Invariances Across Classes?Code1
Inducing Neural Collapse in Imbalanced Learning: Do We Really Need a Learnable Classifier at the End of Deep Neural Network?Code1
EAT: Towards Long-Tailed Out-of-Distribution DetectionCode1
Distribution Alignment: A Unified Framework for Long-tail Visual RecognitionCode1
Class-Conditional Sharpness-Aware Minimization for Deep Long-Tailed RecognitionCode1
Class-Wise Difficulty-Balanced Loss for Solving Class-ImbalanceCode1
Equalization Loss for Long-Tailed Object RecognitionCode1
Escaping Saddle Points for Effective Generalization on Class-Imbalanced DataCode1
Continuous Contrastive Learning for Long-Tailed Semi-Supervised RecognitionCode1
Balanced Meta-Softmax for Long-Tailed Visual RecognitionCode1
Feature Generation for Long-tail ClassificationCode1
Disentangling Label Distribution for Long-tailed Visual RecognitionCode1
FEDIC: Federated Learning on Non-IID and Long-Tailed Data via Calibrated DistillationCode1
Influence-Balanced Loss for Imbalanced Visual ClassificationCode1
LMPT: Prompt Tuning with Class-Specific Embedding Loss for Long-tailed Multi-Label Visual RecognitionCode1
Show:102550
← PrevPage 1 of 3Next →

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1LIFT (ViT-L/14)Top-1 Accuracy82.9—Unverified
2µ2Net+ (ViT-L/16)Top-1 Accuracy82.5—Unverified
3MAM (ViT-B/16)Top-1 Accuracy82.3—Unverified
4LIFT (ViT-B/16)Top-1 Accuracy78.3—Unverified
5VL-LTR (ViT-B-16)Top-1 Accuracy77.2—Unverified
6BALLAD(ResNet-50×16)Top-1 Accuracy76.5—Unverified
7BALLAD(ViT-B-16)Top-1 Accuracy75.7—Unverified
8BALLAD(ResNet-101)Top-1 Accuracy70.5—Unverified
9VL-LTR (ResNet-50)Top-1 Accuracy70.1—Unverified
10BALLAD(ResNet-50)Top-1 Accuracy67.2—Unverified
#ModelMetricClaimedVerifiedStatus
1Cross-Entropy (CE)Error Rate62.75—Unverified
2Cross-Entropy (CE)Error Rate61.68—Unverified
3IBLLossError Rate61.52—Unverified
4Cross-Entropy + Curvature RegularizationError Rate59.5—Unverified
5CE-DRWError Rate58.9—Unverified
6LDAM-DRWError Rate57.96—Unverified
7ELPError Rate57.6—Unverified
8CDB-lossError Rate57.43—Unverified
9CE-DRW-ICError Rate56.9—Unverified
10LDAM-DRW + SSPError Rate56.57—Unverified
#ModelMetricClaimedVerifiedStatus
1RISDAError Rate20.11—Unverified
2Empirical Risk Minimization (ERM, CE)Error Rate13.61—Unverified
3Class-balanced ReweightingError Rate13.46—Unverified
4Class-balanced ResamplingError Rate13.21—Unverified
5IBLLossError Rate12.93—Unverified
6Class-balanced Focal LossError Rate12.9—Unverified
7DecTDEError Rate12.63—Unverified
8M2mError Rate12.5—Unverified
9Prior-LTError Rate12.2—Unverified
10KCLError Rate12—Unverified
#ModelMetricClaimedVerifiedStatus
1LIFT (ViT-L/14@336px)Top-1 Accuracy87.4—Unverified
2LIFT (ViT-L/14)Top-1 Accuracy85.2—Unverified
3GML (ViT-B-16)Top-1 Accuracy82.1—Unverified
4LIFT (ViT-B/16)Top-1 Accuracy80.4—Unverified
5RAC (ViT-B-16)Top-1 Accuracy80.24—Unverified
6GPaCo (2-R152)Top-1 Accuracy79.8—Unverified
7TADE(ResNet-152)Top-1 Accuracy77—Unverified
8ProCo (ResNet50)Top-1 Accuracy75.8—Unverified
9MDCS(Resnet50)Top-1 Accuracy75.6—Unverified
10DeiT-LTTop-1 Accuracy75.1—Unverified