Fine-grained Classification on Aircraft
94.9Top-1 AccCAP
Evaluation results
| Method | Links | |
|---|---|---|
| CAPTraining data setup=primary2021.01 | 94.9 | |
| SLADStudent=ViT-B, Teacher=ViT-L2026.05 | 94.18 | |
| API-NetBackbone=DenseNet-161, Extra Supervision=No2020.02 | 93.9 | |
| Ours (multi scale)Backbone=ResNet-50, Resolution=448, #Parameters=23.9M, scale_mode=multi2019.12 | 93.5 | |
| API-NetBackbone=ResNet-101, Extra Supervision=No2020.02 | 93.4 | |
| LoRAStudent=ViT-B, Teacher=ViT-L2026.05 | 93.25 | |
| API-NetBackbone=ResNet-50, Extra Supervision=No2020.02 | 93 | |
| DCLTraining data setup=primary2021.01 | 93 | |
| Yu et al. 2018bTraining data setup=primary + secondary2021.01 | 92.9 | |
| MC LossTraining data setup=primary2021.01 | 92.9 | |
| GPipeTraining data setup=primary2021.01 | 92.7 | |
| BARMTraining data setup=primary2021.01 | 92.5 | |
| SLADStudent=ViT-S, Teacher=ViT-B2026.05 | 92.5 | |
| SLADStudent=ViT-S, Teacher=ViT-L2026.05 | 92.44 | |
| Ours (single scale)Backbone=ResNet-50, Resolution=448, #Parameters=23.9M, scale_mode=single2019.12 | 92.1 | |
| DFL-CNNExtra Supervision=No2020.02 | 92 | |
| DFLTraining data setup=primary2021.01 | 92 | |
| LoRAStudent=ViT-S, Teacher=ViT-B2026.05 | 91.82 | |
| DFL-CNNBackbone=ResNet-50, Resolution=448, #Parameters=26.3M2019.12 | 91.7 | |
| Deep KSPDExtra Supervision=No2020.02 | 91.5 | |
| LoRAStudent=ViT-S, Teacher=ViT-L2026.05 | 91.46 | |
| NTS-netBackbone=ResNet-50, Resolution=448, #Parameters=25.5M2019.12 | 91.4 | |
| NTS-NetBackbone=ResNet-50, Extra Supervision=No2020.02 | 91.4 | |
| iSQRT-COVBackbone=ResNet-101, Extra Supervision=No2020.02 | 91.4 | |
| Probing*Student=ViT-B, Teacher=ViT-L2026.05 | 91.02 | |
| HBPExtra Supervision=No2020.02 | 90.3 | |
| Probing*Student=ViT-S, Teacher=ViT-L2026.05 | 90.1 | |
| MACNNBackbone=VGG-19, Resolution=448, #Parameters=144M2019.12 | 89.9 | |
| MACNNExtra Supervision=No2020.02 | 89.9 | |
| MaxEntBackbone=ResNet-50, #Parameters=23.9M2019.12 | 89.8 | |
| Grassmann PoolExtra Supervision=No2020.02 | 89.8 | |
| MaxEntExtra Supervision=No2020.02 | 89.8 | |
| Probing*Student=ViT-S, Teacher=ViT-B2026.05 | 89.44 | |
| PCBackbone=ResNet-50, #Parameters=23.9M2019.12 | 89.2 | |
| PCBackbone=DenseNet-161, Extra Supervision=No2020.02 | 89.2 | |
| ProbingStudent=ViT-S, Teacher=ViT-L2026.05 | 89.2 | |
| G2DeNetBackbone=VGGNet-16, Extra Supervision=No2020.02 | 89 | |
| BoTBackbone=VGGNet-16, Extra Supervision=Yes2020.02 | 88.4 | |
| RACNNBackbone=VGGNet-19, Extra Supervision=No2020.02 | 88.4 | |
| RACNNBackbone=VGG-19, Resolution=448, #Parameters=129M2019.12 | 88.2 | |
| w/ LaViDStudent=MNV2, Teacher=QN+R502026.06 | 88.08 | |
| LSStudent=MNV2, Teacher=RN-502026.06 | 87.96 | |
| LRBPExtra Supervision=No2020.02 | 87.3 | |
| KPExtra Supervision=No2020.02 | 86.9 | |
| MG-CNNBackbone=VGGNet-19, Extra Supervision=Yes2020.02 | 86.6 | |
| MLKDStudent=MNV2, Teacher=RN-502026.06 | 86.38 | |
| w/ LaViDStudent=RN-18, Teacher=QN+R502026.06 | 86.21 | |
| LaViDStudent=MNV2, Teacher=Qwen2026.06 | 86.21 | |
| LSStudent=RN-18, Teacher=RN-502026.06 | 86.02 | |
| KDStudent=MNV2, Teacher=RN-502026.06 | 85.48 | |
| FitNetStudent=MNV2, Teacher=LLaVA2026.06 | 85.46 | |
| DKDStudent=MNV2, Teacher=RN-502026.06 | 85.39 | |
| MLKDStudent=RN-18, Teacher=RN-502026.06 | 85.35 | |
| Ind StudentStudent=MNV2, Teacher=None2026.06 | 85.27 | |
| DKDStudent=RN-18, Teacher=RN-502026.06 | 84.48 | |
| MLKDStudent=SNV2, Teacher=RN-502026.06 | 84.12 | |
| B-CNNExtra Supervision=No2020.02 | 84.1 | |
| LSStudent=SNV2, Teacher=RN-502026.06 | 84.09 | |
| MaKDStudent=MNV2, Teacher=InternVL2026.06 | 83.81 | |
| w/ LaViDStudent=SNV2, Teacher=QN+R502026.06 | 83.72 | |
| LaViDStudent=RN-18, Teacher=Qwen2026.06 | 83.22 | |
| RKDStudent=MNV2, Teacher=RN-502026.06 | 82.98 | |
| DKDStudent=SNV2, Teacher=RN-502026.06 | 82.63 | |
| CRDStudent=MNV2, Teacher=LLaVA2026.06 | 82.31 | |
| LaViDStudent=SNV2, Teacher=Qwen2026.06 | 81.63 | |
| KDStudent=RN-18, Teacher=RN-502026.06 | 81.16 | |
| KDStudent=SNV2, Teacher=RN-502026.06 | 80.85 | |
| CRDStudent=RN-18, Teacher=LLaVA2026.06 | 80.81 | |
| Ind StudentStudent=SNV2, Teacher=None2026.06 | 80.65 | |
| FitNetStudent=SNV2, Teacher=LLaVA2026.06 | 80.61 | |
| MaKDStudent=RN-18, Teacher=InternVL2026.06 | 80.45 | |
| FitNetStudent=RN-18, Teacher=LLaVA2026.06 | 79.95 | |
| MaKDStudent=SNV2, Teacher=InternVL2026.06 | 79 | |
| Ind StudentStudent=RN-18, Teacher=None2026.06 | 78.95 | |
| CRDStudent=SNV2, Teacher=LLaVA2026.06 | 78.78 | |
| RKDStudent=RN-18, Teacher=RN-502026.06 | 78.18 | |
| RKDStudent=SNV2, Teacher=RN-502026.06 | 74.56 | |
| Claude 3.5 sonnetOptimization strategy=false2025.12 | 64.4 | |
| GPT-4oOptimization strategy=false2025.12 | 63.8 | |
| Step 1vOptimization strategy=false2025.12 | 51.8 | |
| Gemini 1.5 proOptimization strategy=false2025.12 | 50.8 | |
| Doubao 1.5 vision proOptimization strategy=false2025.12 | 49.4 | |
| GLM v plusOptimization strategy=false2025.12 | 42.6 | |
| Qwen-VL-chat-78BOptimization strategy=false2025.12 | 35.8 | |
| Fair Context Learning (FCL)mode=Test-Time Adaptation2026.02 | 34.17 | |
| InternVL 2.5-8BOptimization strategy=true2025.12 | 31.8 | |
| APMPre-trained on ImageNet=false2024.10 | 29.7 | |
| Hunyuan visionOptimization strategy=false2025.12 | 29.6 | |
| LLaVA 1.5-7BOptimization strategy=true2025.12 | 25.6 | |
| MTAmode=Test-Time Adaptation2026.02 | 24.84 | |
| ZEROmode=Test-Time Adaptation2026.02 | 24.81 | |
| TPTPre-trained on ImageNet=false2024.10 | 24.8 | |
| TPSmode=Test-Time Adaptation2026.02 | 24.78 | |
| TTLmode=Test-Time Adaptation2026.02 | 24.75 | |
| R-TPTmode=Test-Time Adaptation2026.02 | 24.03 | |
| C-TPTmode=Test-Time Adaptation2026.02 | 23.94 | |
| CLIPmode=Zero-shot2026.02 | 23.91 | |
| CLIP-VIT-B/16Pre-trained on ImageNet=false2024.10 | 23.7 | |
| TPTmode=Test-Time Adaptation2026.02 | 23.55 | |
| EnsemblePre-trained on ImageNet=false2024.10 | 23.2 |