Image Classification on Flowers (Top-1 Error)
3.44Top-1 ErrorIndividual Networks
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Individual NetworksBackbone=ResNet-502018.01 | 3.44 | — | |
| Individual NetworksBackbone=DenseNet-1212018.01 | 3.49 | — | |
| Individual NetworksBackbone=VGG-16 BN2018.01 | 4.55 | — | |
| PiggybackBackbone=VGG-16 BN2018.01 | 4.84 | — | |
| PiggybackBackbone=ResNet-502018.01 | 5.23 | — | |
| PiggybackBackbone=DenseNet-1212018.01 | 5.31 | — | |
| PackNetBackbone=ResNet-50, Task addition order=↓2018.01 | 6.96 | — | |
| PiggybackBackbone=VGG-16, # Models=1, Size=621 MB2018.01 | 7.19 | — | |
| Individual NetworksBackbone=VGG-16, # Models=6, Size=3,222 MB2018.01 | 7.35 | — | |
| PackNetBackbone=VGG-16 BN, Task addition order=↓2018.01 | 7.82 | — | |
| PackNetBackbone=VGG-16 BN, Task addition order=↑2018.01 | 7.85 | — | |
| PackNetBackbone=DenseNet-121, Task addition order=↑2018.01 | 8.46 | — | |
| PackNetBackbone=DenseNet-121, Task addition order=↓2018.01 | 8.94 | — | |
| PackNetBackbone=ResNet-50, Task addition order=↑2018.01 | 9.45 | — | |
| PackNetBackbone=VGG-16, # Models=1, Size=587 MB, Task addition order=Sketch -> CUBS2018.01 | 10.25 | — | |
| PackNetBackbone=VGG-16, # Models=1, Size=587 MB, Task addition order=CUBS -> Sketch2018.01 | 10.33 | — | |
| Classifier OnlyBackbone=ResNet-502018.01 | 14.01 | — | |
| Classifier OnlyBackbone=DenseNet-1212018.01 | 16.56 | — | |
| Classifier OnlyBackbone=VGG-16 BN2018.01 | 19.38 | — | |
| Classifier OnlyBackbone=VGG-16, # Models=1, Size=537 MB2018.01 | 20.01 | — | |
| EfficientNet-L2Backbone=EfficientNet-L2, SAM Optimization=true, Pre-training Dataset=ImageNet plus JFT2020.10 | 35 | — | |
| EfficientNet2020.10 | 37 | — | |
| EfficientNet-L2Backbone=EfficientNet-L2, SAM Optimization=false, Pre-training Dataset=ImageNet plus JFT2020.10 | 40 | — | |
| EfficientNet-b7Backbone=EfficientNet-b7, SAM Optimization=true, Pre-training Dataset=ImageNet-only2020.10 | 63 | — | |
| BiT-MPre-training Dataset=ImageNet-only2020.10 | 70 | — | |
| EfficientNet-b7Backbone=EfficientNet-b7, SAM Optimization=false, Pre-training Dataset=ImageNet-only2020.10 | 116 | — |