Image Classification on CIFAR-10 LT (Accuracy by Sampling Ratio) benchmark leaderboard · SOTA2 Research

Image Classification on CIFAR-10 LT (Accuracy by Sampling Ratio)

91.24Acc (Ratio 10)Dynamic Loss

Evaluation Results

Method
Dynamic Loss2022.11	91.24	88.3	86.46	82.95	87.24
Dynamic Loss + Logit Adjustment2022.11	91.1	89.54	85.52	83.43	87.4
Balanced-Softmax2022.11	91.01	88.85	86.44	82.31	87.15
Dynamic Loss + Balanced-Softmax2022.11	90.99	89.66	85.49	83.21	87.34
MiSLAS2022.11	90	88.52	85.7	82.1	86.58
WD2022.11	89.8	84.81	79.66	74.84	82.28
Logit Adjustment2022.11	89.64	86.77	82.61	78.38	84.35
FaMUS2022.11	87.9	86.24	83.32	80.96	84.61
LDAM-DRW2022.11	87.68	85.51	81.64	78.02	83.21
CB Focal2022.11	87.49	84.36	79.27	74.57	81.42
Focal Loss2022.11	86.66	82.76	76.71	70.38	79.13
Cross entropy2022.11	86.39	82.23	74.81	70.36	78.45
CMO2022.11	83.26	89.27	87.19	85.35	86.27