Calibration on ImageNet long-tailed (test)
1.5063ECEMIR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MIRBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.5063 | 0.3217 | 1.531 | 3.6285 | |
| MIRBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 1.6207 | 0.3205 | 1.6562 | 3.7481 | |
| PTSBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.73 | 0.316 | 1.7401 | 3.9018 | |
| TSBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.8541 | 0.3956 | 1.7472 | 3.8478 | |
| PTSKBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.8558 | 0.3475 | 1.9875 | 3.9175 | |
| ETSBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.8642 | 0.3926 | 1.7622 | 3.852 | |
| confBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 1.8724 | 0.338 | 2.0168 | 3.9776 | |
| TSBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 2.1965 | 0.31 | 2.058 | 3.8407 | |
| ETSBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 2.2704 | 0.3264 | 2.0648 | 4.0024 | |
| PTSBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 5.0023 | 0.6512 | 4.9927 | 7.6502 | |
| confBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 7.4933 | 0.8672 | 7.461 | 7.9229 | |
| IRBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 8.0541 | 1.0642 | 7.9947 | 8.8631 | |
| IRBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 8.5778 | 1.0508 | 8.5546 | 9.3678 | |
| PTSKBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 9.1862 | 1.7478 | 9.208 | 11.4242 | |
| HBBackbone=ResNet50, Classifier re-training=[18], Framework=ProCal2023.06 | 12.9257 | 7.567 | 13.4697 | 15.3259 | |
| HBBackbone=ResNet50, Classifier re-training=[18], Framework=Base2023.06 | 13.4644 | 7.5672 | 13.5931 | 16.2915 |