Image Classification on ImageNet (Accuracy and Calibration)
77.5AccuracyCVP
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| CVPRel. Time=2.9×, Backbone=ResNet-502026.06 | 77.5 | 6.18 | 10.8 | 23.8 | 0.86 | 1.4 | 0.31 | |
| Mean NetworkRel. Time=1.0×, Backbone=ResNet-502026.06 | 77.2 | 6.26 | 7.1 | 23.7 | 0.89 | 3.48 | 0.32 | |
| Mean Network + Temp. Sc.Rel. Time=1.0×, Backbone=ResNet-502026.06 | 77.2 | 6.28 | 10.2 | 23.5 | 0.88 | 1.34 | 0.32 | |
| StreamliningRel. Time=2.6×, Backbone=ResNet-502026.06 | 77.2 | 6.28 | 10.1 | 23.3 | 0.88 | 1.43 | 0.32 | |
| MC Sampling (4 samples)Rel. Time=4.0×, Backbone=ResNet-502026.06 | 76.9 | 6.49 | 10.3 | 23 | 0.89 | 1.6 | 0.32 | |
| MC Sampling (2 samples)Rel. Time=2.0×, Backbone=ResNet-502026.06 | 76.4 | 6.69 | 9.6 | 22.2 | 0.91 | 1.18 | 0.33 |