Image Classification on iNaturalist 2018 (val)
84.1Top-1 AccuracyOMNIVORE (Swin-L)
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| OMNIVORE (Swin-L)Backbone=Swin-L, Resolution=384x3842022.01 | 84.1 | — | — | — | — | — | — | — | — | — | |
| OMNIVORE (Swin-B)Backbone=Swin-B, Resolution=384x3842022.01 | 82.6 | — | — | — | — | — | — | — | — | — | |
| ViT-L/16Backbone=ViT-L/16, Resolution=384x3842022.01 | 81.7 | — | — | — | — | — | — | — | — | — | |
| EfficientNet B8Backbone=EfficientNet B82022.01 | 81.3 | — | — | — | — | — | — | — | — | — | |
| EfficientNet B7Backbone=EfficientNet B72022.01 | 80.6 | — | — | — | — | — | — | — | — | — | |
| ViT-B/16Backbone=ViT-B/16, Resolution=384x3842022.01 | 79.8 | — | — | — | — | — | — | — | — | — | |
| DeiT-BBackbone=DeiT-B, Resolution=384x3842022.01 | 79.5 | — | — | — | — | — | — | — | — | — | |
| EfficientNet B6Backbone=EfficientNet B62022.01 | 79.1 | — | — | — | — | — | — | — | — | — | |
| OMNIVORE (Swin-L)Backbone=Swin-L, Resolution=224x2242022.01 | 78 | — | — | — | — | — | — | — | — | — | |
| OMNIVORE (Swin-B)Backbone=Swin-B, Resolution=224x2242022.01 | 76.3 | — | — | — | — | — | — | — | — | — | |
| XCiT-L24/16Architecture=XCiT-L24/16, distillation=true2021.06 | 75.6 | — | — | — | — | — | — | — | — | — | |
| DeiT III-LBackbone=ViT-L, Pre-training=ImageNet-1k, Epochs=400, Resolution=224x224, Crop ratio=1.02022.04 | 75.6 | — | — | — | — | — | — | — | — | — | |
| iBOTBackbone=ViT-B/16, Evaluation Protocol=Fine-tuning2021.11 | 74.6 | — | — | — | — | — | — | — | — | — | |
| DeiT-B/16Architecture=DeiT-B/16, distillation=true2021.06 | 73.7 | — | — | — | — | — | — | — | — | — | |
| iBOTBackbone=ViT-S/16, Evaluation Protocol=Fine-tuning2021.11 | 73.7 | — | — | — | — | — | — | — | — | — | |
| DeiT III-BBackbone=ViT-B, Pre-training=ImageNet-1k, Epochs=400, Resolution=224x224, Crop ratio=1.02022.04 | 73.6 | — | — | — | — | — | — | — | — | — | |
| MBJ + RIDE (4 experts)Epochs=200, Backbone=ResNet-502020.08 | 73.2 | — | — | — | — | — | — | — | — | — | |
| Rand.Backbone=ViT-B/16, Evaluation Protocol=Fine-tuning2021.11 | 73.2 | — | — | — | — | — | — | — | — | — | |
| DeiT-BPre-training=ImageNet-1k, Crop ratio=0.8752022.04 | 73.2 | — | — | — | — | — | — | — | — | — | |
| XCiT-M24/16Architecture=XCiT-M24/16, distillation=true2021.06 | 72.6 | — | — | — | — | — | — | — | — | — | |
| RIDE (4 experts)Epochs=200, Backbone=ResNet-502020.08 | 72.6 | — | — | — | — | — | — | — | — | — | |
| DINOBackbone=ViT-B/16, Evaluation Protocol=Fine-tuning2021.11 | 72.6 | — | — | — | — | — | — | — | — | — | |
| BEITBackbone=ViT-B/16, Evaluation Protocol=Fine-tuning2021.11 | 72.3 | — | — | — | — | — | — | — | — | — | |
| DINOBackbone=ViT-S/16, Evaluation Protocol=Fine-tuning2021.11 | 72 | — | — | — | — | — | — | — | — | — | |
| BCLBackbone=ResNet-50, Training Epochs=1002022.07 | 71.8 | — | — | — | — | — | — | — | — | — | |
| RIDE(2 experts)Backbone=ResNet-50, Training Epochs=1002022.07 | 71.4 | — | — | — | — | — | — | — | — | — | |
| Rand.Backbone=ViT-S/16, Evaluation Protocol=Fine-tuning2021.11 | 70.7 | — | — | — | — | — | — | — | — | — | |
| SAMixPBackbone=ResNeXt-1012021.11 | 70.56 | — | — | — | — | — | — | — | — | — | |
| SAMixBackbone=ResNeXt-1012021.11 | 70.54 | — | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=ResNeXt-1012021.11 | 70.49 | — | — | — | — | — | — | — | — | — | |
| PuzzleMixBackbone=ResNeXt-1012021.11 | 70.12 | — | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=ResNeXt-1012021.11 | 70.01 | — | — | — | — | — | — | — | — | — | |
| MBJEpochs=200, Backbone=ResNet-502020.08 | 70 | — | — | — | — | — | — | — | — | — | |
| Grafit ResNet-50 [52]Pre-training=ImageNet-1k, Crop ratio=0.8752022.04 | 69.8 | — | — | — | — | — | — | — | — | — | |
| CutMixBackbone=ResNeXt-1012021.11 | 69.75 | — | — | — | — | — | — | — | — | — | |
| BBNEpochs=200, Backbone=ResNet-502020.08 | 69.7 | — | — | — | — | — | — | — | — | — | |
| Decouple-LWSEpochs=200, Backbone=ResNet-502020.08 | 69.5 | — | — | — | — | — | — | — | — | — | |
| DisAlignBackbone=ResNet-50, Training Epochs=1002022.07 | 69.5 | — | — | — | — | — | — | — | — | — | |
| FMixBackbone=ResNeXt-1012021.11 | 69.46 | — | — | — | — | — | — | — | — | — | |
| Decouple-τ-normEpochs=200, Backbone=ResNet-502020.08 | 69.3 | — | — | — | — | — | — | — | — | — | |
| ManifoldMixBackbone=ResNeXt-1012021.11 | 69.3 | — | — | — | — | — | — | — | — | — | |
| ResizeMixBackbone=ResNeXt-1012021.11 | 69.3 | — | — | — | — | — | — | — | — | — | |
| UniMix+BayiasE2E=true, Backbone=ResNet-502021.11 | 69.15 | — | — | — | 8.27 | — | — | — | — | — | |
| ResNet-152 [4]Pre-training=ImageNet-1k, Crop ratio=0.8752022.04 | 69.1 | — | — | — | — | — | — | — | — | — | |
| Feature Space AugmentationBackbone=ResNet-1522020.08 | 69.08 | — | — | — | — | — | — | — | — | — | |
| XCiT-S24/16Architecture=XCiT-S24/16, distillation=true2021.06 | 68.8 | — | — | — | — | — | — | — | — | — | |
| KCLBackbone=ResNet-50, Training Epochs=1002022.07 | 68.6 | — | — | — | — | — | — | — | — | — | |
| BEITBackbone=ViT-S/16, Evaluation Protocol=Fine-tuning2021.11 | 68.5 | — | — | — | — | — | — | — | — | — | |
| Feature Space AugmentationBackbone=ResNet-1012020.08 | 68.39 | — | — | — | — | — | — | — | — | — | |
| Hybrid-PSCBackbone=ResNet-50, Training Epochs=1002022.07 | 68.1 | — | — | — | — | — | — | — | — | — | |
| SSPBackbone=ResNet-50, Training Epochs=1002022.07 | 68.1 | — | — | — | — | — | — | — | — | — | |
| LDAM+DRWE2E=false, Backbone=ResNet-502021.11 | 68 | — | — | — | 7.12 | — | — | — | — | — | |
| Decouple-cRTEpochs=200, Backbone=ResNet-502020.08 | 67.6 | — | — | — | — | — | — | — | — | — | |
| c-RTE2E=false, Backbone=ResNet-502021.11 | 67.6 | — | — | — | 6.72 | — | — | — | — | — | |
| MixupBackbone=ResNeXt-1012021.11 | 67.56 | — | — | — | — | — | — | — | — | — | |
| DeiT III-SBackbone=ViT-S, Pre-training=ImageNet-1k, Epochs=400, Resolution=224x224, Crop ratio=1.02022.04 | 67.1 | — | — | — | — | — | — | — | — | — | |
| IEM*Epochs=90, Backbone=ResNet-502020.08 | 67 | — | — | — | — | — | — | — | — | — | |
| CBBackbone=ResNet-1522020.08 | 66.97 | — | — | — | — | — | — | — | — | — | |
| VanillaBackbone=ResNeXt-1012021.11 | 66.94 | — | — | — | — | — | — | — | — | — | |
| MBJEpochs=90, Backbone=ResNet-502020.08 | 66.9 | — | — | — | — | — | — | — | — | — | |
| Hybrid-SCBackbone=ResNet-50, Training Epochs=1002022.07 | 66.7 | — | — | — | — | — | — | — | — | — | |
| BBNEpochs=90, Backbone=ResNet-502020.08 | 66.3 | — | — | — | — | — | — | — | — | — | |
| BBNBackbone=ResNet-50, Training Epochs=1002022.07 | 66.3 | — | — | — | — | — | — | — | — | — | |
| BBNE2E=false, Backbone=ResNet-502021.11 | 66.29 | — | — | — | 5.41 | — | — | — | — | — | |
| BaselineBackbone=ResNet-1522020.08 | 66.17 | — | — | — | — | — | — | — | — | — | |
| LDAM-DRWEpochs=200, Backbone=ResNet-502020.08 | 66.1 | — | — | — | — | — | — | — | — | — | |
| Feature Space AugmentationBackbone=ResNet-502020.08 | 65.91 | — | — | — | — | — | — | — | — | — | |
| FSAEpochs=90, Backbone=ResNet-502020.08 | 65.9 | — | — | — | — | — | — | — | — | — | |
| Decouple-LWSEpochs=90, Backbone=ResNet-502020.08 | 65.9 | — | — | — | — | — | — | — | — | — | |
| LWSBackbone=ResNet-50, Training Epochs=1002022.07 | 65.9 | — | — | — | — | — | — | — | — | — | |
| Decouple-τ-normEpochs=90, Backbone=ResNet-502020.08 | 65.6 | — | — | — | — | — | — | — | — | — | |
| τ-normBackbone=ResNet-50, Training Epochs=1002022.07 | 65.6 | — | — | — | — | — | — | — | — | — | |
| IBBackbone=ResNet-502021.10 | 65.39 | — | — | 84.98 | — | — | — | — | — | — | |
| Basel.(CE)Epochs=200, Backbone=ResNet-502020.08 | 65.3 | — | — | — | — | — | — | — | — | — | |
| Decouple-cRTEpochs=90, Backbone=ResNet-502020.08 | 65.2 | — | — | — | — | — | — | — | — | — | |
| cRTBackbone=ResNet-50, Training Epochs=1002022.07 | 65.2 | — | — | — | — | — | — | — | — | — | |
| BaselineBackbone=ResNet-1012020.08 | 65.19 | — | — | — | — | — | — | — | — | — | |
| SAMixPBackbone=ResNet-502021.11 | 65.16 | — | — | — | — | — | — | — | — | — | |
| SAMixBackbone=ResNet-502021.11 | 64.84 | — | — | — | — | — | — | — | — | — | |
| AutoMixBackbone=ResNet-502021.11 | 64.73 | — | — | — | — | — | — | — | — | — | |
| LDAMEpochs=90, Backbone=ResNet-502020.08 | 64.6 | — | — | — | — | — | — | — | — | — | |
| LDAM-DRWEpochs=90, Backbone=ResNet-502020.08 | 64.6 | — | — | — | — | — | — | — | — | — | |
| LDAMBackbone=ResNet-502021.10 | 64.58 | — | — | 83.52 | — | — | — | — | — | — | |
| LDAME2E=true, Backbone=ResNet-502021.11 | 64.58 | — | — | — | 3.7 | — | — | — | — | — | |
| CE+SSPEpochs=90, Backbone=ResNet-502020.08 | 64.4 | — | — | — | — | — | — | — | — | — | |
| PuzzleMixBackbone=ResNet-502021.11 | 64.36 | — | — | — | — | — | — | — | — | — | |
| SaliencyMixBackbone=ResNet-502021.11 | 64.27 | — | — | — | — | — | — | — | — | — | |
| ResizeMixBackbone=ResNet-502021.11 | 64.12 | — | — | — | — | — | — | — | — | — | |
| CutMixBackbone=ResNet-502021.11 | 63.91 | — | — | — | — | — | — | — | — | — | |
| OLTRE2E=false, Backbone=ResNet-502021.11 | 63.9 | — | — | — | 3.02 | — | — | — | — | — | |
| CBBackbone=ResNet-101, Source=Literature2020.08 | 63.88 | — | — | — | — | — | — | — | — | — | |
| FMixBackbone=ResNet-502021.11 | 63.71 | — | — | — | — | — | — | — | — | — | |
| CB-CEE2E=true, Backbone=ResNet-502021.11 | 63.5 | — | — | — | 2.62 | — | — | — | — | — | |
| ManifoldMixBackbone=ResNet-502021.11 | 63.46 | — | — | — | — | — | — | — | — | — | |
| Decouple-NCMEpochs=200, Backbone=ResNet-502020.08 | 63.1 | — | — | — | — | — | — | — | — | — | |
| MixupBackbone=ResNet-502021.11 | 62.69 | — | — | — | — | — | — | — | — | — | |
| VanillaBackbone=ResNet-502021.11 | 62.53 | — | — | — | — | — | — | — | — | — | |
| BaselineBackbone=ResNet-502020.08 | 62.27 | — | — | — | — | — | — | — | — | — | |
| CBBackbone=ResNet-50, Source=Literature2020.08 | 61.12 | — | — | — | — | — | — | — | — | — | |
| CBBackbone=ResNet-502021.10 | 61.12 | — | — | 81.03 | — | — | — | — | — | — |