Difficulty-Net (ResNet-50 using RandAugment, single model)

Reported on 5 benchmarks across 5 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Methodology3 results

Generalized Few-Shot ClassificationonImageNet-LT
Top-1 Accuracy· 2022-09-07
57.4
best: 82.9 (LIFT (ViT-L/14))
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition arXiv:2209.02960
Long-tail LearningonImageNet-LT
Top-1 Accuracy· 2022-09-07
57.4
best: 82.9 (LIFT (ViT-L/14))
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition arXiv:2209.02960
Generalized Few-Shot LearningonImageNet-LT
Top-1 Accuracy· 2022-09-07
57.4
best: 82.9 (LIFT (ViT-L/14))
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition arXiv:2209.02960

Computer Vision2 results

Image ClassificationonImageNet-LT
Top-1 Accuracy· 2022-09-07
57.4
best: 82.9 (LIFT (ViT-L/14))
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition arXiv:2209.02960
Few-Shot Image ClassificationonImageNet-LT
Top-1 Accuracy· 2022-09-07
57.4
best: 82.9 (LIFT (ViT-L/14))
Difficulty-Net: Learning to Predict Difficulty for Long-Tailed Recognition arXiv:2209.02960