AI Paper Atlas

Research map / Benchmarks / ImageNet-LT

ImageNet-LT leaderboard

52 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on ImageNet-LT. Best accuracy: 79.17 (Ours, NeurIPS 2025).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

ImageNet-LT · accuracy

40 papers, each with its best reported number for its own method; higher is better.

#accuracy ↑MethodPaperTopic
179.17Ours
ViT-B/16
Salient Concept-Aware Generative Data Augmentation
NeurIPS 2025 · Table 3
Text-to-image diffusion
278.9NAO-centroid
ResNet-50
Norm-guided latent space exploration for text-to-image generation
NeurIPS 2023 · Few-shot and long-tail results
Text-to-image diffusion
377.4HGLTR
ViT-B/16
HGLTR: Hierarchical Knowledge Injection for Calibrating Pre-trained Models in Long-Tail Recognition
AAAI 2026 · Table 2
Recognition and long-tail
476.33DGN
ViT-Base (LIFT)
Dynamic Group Normalization: Spatio-Temporal Adaptation to Evolving Data Statistics
CVPR 2025 · Table 5
RNNs and normalization
563.8LiVT (384 resolution)
ViT-Base-16
Learning Imbalanced Data With Vision Transformers
CVPR 2023 · Table: Top-1 accuracy of ResNe
Attention and transformers
663TCL
ResNeXt-50
Trust-calibrated Collaborative Learning for Long-Tailed Visual Recognition
CVPR 2026 · Table 3
Recognition and long-tail
762.9VICAL
ResNeXt-50
VICAL: Vicinal Consistency Alignment for Long-Tailed Visual Recognition
ECCV 2026 · Table 1
Recognition and long-tail
861.8LiVT + Ours
LiVT
Leverage Class-Specific Accuracy to Guide Data Generation for Improving Image Classification
ICML 2024 · Table 4
Autoregressive and latent generation
961.1LGLA (GC)
ResNeXt-50
Local and Global Logit Adjustments for Long-Tailed Learning
ICCV 2023 · Table 2
Recognition and long-tail
1060.8PRLBreaking Long-Tailed Learning Bottlenecks: A Controllable Paradigm with Hypernetwork-Generated Diverse Experts
NeurIPS 2024 · Table 1
Recognition and long-tail
30 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

ImageNet-LT · overall accuracy

13 papers, each with its best reported number for its own method; higher is better.

3 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.