Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MedMnist

Benchmarks

Task NameDataset NameSOTA ResultTrend
Medical Image ClassificationMedMnist BloodMnist (test)
Accuracy99
65
Medical Image ClassificationMedMNIST Breast (test)
Accuracy92.8
36
Medical Image ClassificationMedMNIST Derma (test)
Accuracy84.1
36
Medical Image ClassificationMedMNIST Pneumonia (test)
Accuracy96.2
36
3D Image ClassificationMedMNIST 3D v2 (test)
Organ Accuracy0.998
36
Medical Image ClassificationMedMNIST OrganS (test)
Accuracy82.4
30
Supervised Learning3D MedMNIST
Top-1 Balanced Accuracy77.14
24
Supervised LearningMedMNIST 2D
Top-1 Balanced Accuracy82.42
24
Medical Image ClassificationMedMNIST DermaMNIST v2 (test)
Accuracy94.4
22
Image ClassificationMedMNIST OrganAMNIST (test)
Accuracy97.62
22
Semi-supervised classificationMedMNIST Breast (10% train 90% test)
Accuracy85.7
20
Base-to-New GeneralizationMedMNIST+ Base-to-New v2
Base Accuracy57.84
20
Image ClassificationMedMNIST 2D OrganAMNIST
Accuracy96.2
20
Image ClassificationMedMNIST Sub
Final Epoch Accuracy73.61
18
Adversarial AttackChestMNIST (test)
KMRa0.03
15
Image ClassificationMEDMNISTC (Dir(0.1))
Accuracy66.97
13
Image ClassificationMEDMNISTA Dir(0.5)
Accuracy41.22
13
Image ClassificationMedMNIST
DermalMNIST Top-1 Accuracy82.1
13
Medical Image ClassificationMedMNIST v2 (test)
Average Accuracy26.8
13
Image ClassificationMedMNIST 2D OrganCMNIST
Accuracy94.4
12
Image ClassificationMedMNIST BloodMNIST 2D
Accuracy98.7
12
Image ClassificationMedMNIST PneumoniaMNIST 2D
Accuracy93.6
12
Image ClassificationMedMNIST 2D OCTMNIST
Accuracy92.2
12
Image ClassificationMedMNIST PathMNIST (test)
Accuracy86.63
12
ClassificationMedMNIST3D Nodule (test)
Accuracy88.7
11
Showing 25 of 52 rows