Domain-Incremental Learning on CORe50
99.5Avg Accuracy (A)Residual SODAP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Residual SODAPCL-Type=Ours2026.03 | 99.5 | — | — | 0.3 | |
| DER++CL-Type=Reh-CL2026.03 | 99.4 | — | — | 6.1 | |
| Online EWCCL-Type=Reg-CL2026.03 | 98.9 | — | — | 2.9 | |
| OS-Prompt++CL-Type=PCL2026.03 | 98.3 | — | — | 1.4 | |
| OS-PromptCL-Type=PCL2026.03 | 98.2 | — | — | 1.2 | |
| Dual-promptCL-Type=PCL2026.03 | 97.5 | — | — | 3.6 | |
| Coda-PromptCL-Type=PCL2026.03 | 97.4 | — | — | 5.6 | |
| CVLCBackbone=CLIP, Shots=8-shot2026.06 | 96.96 | 96.08 | — | — | |
| CVLCBackbone=CLIP, Shots=4-shot2026.06 | 96.44 | 95.43 | — | — | |
| CVLCBackbone=CLIP, Shots=2-shot2026.06 | 96.21 | 95.11 | — | — | |
| CVLCBackbone=CLIP, Shots=1-shot2026.06 | 95.74 | 94.35 | — | — | |
| CoVONModel=Vision Transformer (ViT-B/16)2026.06 | 95.14 | — | — | — | |
| MoE-AdaptersCL-Type=Arch-CL2026.03 | 94.9 | — | — | 4.3 | |
| MIRAScenario=Domain Incremental Learning2025.11 | 93.89 | — | 0 | — | |
| CoVON-NoMModel=Vision Transformer (ViT-B/16)2026.06 | 93.84 | — | — | — | |
| CoVONModel=CLIP with pretrained ViT-B/16 as vision encoder2026.06 | 92.64 | — | — | — | |
| Disentangled Weight FusionBuffer=0/class2026.04 | 92.36 | 93.57 | — | — | |
| MoP-CLIPModel=CLIP with pretrained ViT-B/16 as vision encoder2026.06 | 92.29 | — | — | — | |
| DUCTBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 91.95 | 94.47 | — | — | |
| DUCTBuffer=0/class2026.04 | 91.95 | 94.47 | — | — | |
| S-liPromptsModel=CLIP with pretrained ViT-B/16 as vision encoder2026.06 | 90.1 | — | — | — | |
| ICONScenario=Domain Incremental Learning2025.11 | 89.01 | — | 0.17 | — | |
| PGO-BEnBackbone=CLIP, Shots=8-shot2026.06 | 88.43 | 85.34 | — | — | |
| CODA-PromptBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 87.92 | 91.57 | — | — | |
| CODA-PromptBuffer=0/class2026.04 | 87.92 | 91.57 | — | — | |
| PINABuffer=0/class2026.04 | 87.38 | 89.82 | — | — | |
| PINAModel=CLIP with pretrained ViT-B/16 as vision encoder2026.06 | 87.38 | — | — | — | |
| PGO-BEnBackbone=CLIP, Shots=4-shot2026.06 | 87.38 | 83.81 | — | — | |
| PINAModel=Vision Transformer (ViT-B/16)2026.06 | 87.26 | — | — | — | |
| PGO-BEnBackbone=CLIP, Shots=2-shot2026.06 | 86.73 | 82.81 | — | — | |
| S-PromptsScenario=Domain Incremental Learning2025.11 | 86.5 | — | 0.92 | — | |
| EASEBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 86.3 | 87.02 | — | — | |
| ReplayBackbone=ViT-B/16-IN1K, Exemplars=10 per class2024.10 | 85.56 | 92.21 | — | — | |
| EWC*Model=Vision Transformer (ViT-B/16)2026.06 | 85.35 | — | — | — | |
| CODA-PModel=Vision Transformer (ViT-B/16)2026.06 | 85.31 | — | — | — | |
| DualPromptBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 84.53 | 87.27 | — | — | |
| DualPromptBuffer=0/class2026.04 | 84.53 | 87.27 | — | — | |
| CODA-PScenario=Domain Incremental Learning2025.11 | 84.36 | — | 0.64 | — | |
| CP-PromptBackbone=CLIP, Shots=8-shot2026.06 | 84.08 | 82.79 | — | — | |
| L2PBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 83.57 | 87.87 | — | — | |
| L2PBuffer=0/class2026.04 | 83.57 | 87.87 | — | — | |
| S-iPromptsModel=Vision Transformer (ViT-B/16)2026.06 | 83.5 | — | — | — | |
| PGO-BEnBackbone=CLIP, Shots=1-shot2026.06 | 83.19 | 78.14 | — | — | |
| LAEScenario=Domain Incremental Learning2025.11 | 83.09 | — | 0.17 | — | |
| CP-PromptBackbone=CLIP, Shots=4-shot2026.06 | 82.27 | 81.26 | — | — | |
| S-iPromptBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 81.95 | 83.38 | — | — | |
| S-iPromptBuffer=0/class2026.04 | 81.95 | 83.38 | — | — | |
| CP-PromptBackbone=CLIP, Shots=2-shot2026.06 | 81.65 | 78.68 | — | — | |
| DualPromptScenario=Domain Incremental Learning2025.11 | 81.41 | — | 0.21 | — | |
| L2PBuffer=50/class2026.04 | 81.07 | 82.72 | — | — | |
| L2P*Backbone=CLIP, Shots=2-shot2026.06 | 80.94 | 79.55 | — | — | |
| L2PScenario=Domain Incremental Learning2025.11 | 80.72 | — | 0.51 | — | |
| DualPromptModel=Vision Transformer (ViT-B/16)2026.06 | 80.68 | — | — | — | |
| L2P*Backbone=CLIP, Shots=1-shot2026.06 | 80.19 | 77.85 | — | — | |
| S-PromptBackbone=CLIP, Shots=8-shot2026.06 | 80.13 | 78.77 | — | — | |
| ERBuffer=50/class2026.04 | 80.1 | 82.06 | — | — | |
| DER++Buffer=50/class2026.04 | 79.7 | 82.01 | — | — | |
| S-PromptBackbone=CLIP, Shots=4-shot2026.06 | 79.63 | 77.26 | — | — | |
| S-PromptBackbone=CLIP, Shots=2-shot2026.06 | 79.53 | 74.95 | — | — | |
| L2P*Backbone=CLIP, Shots=4-shot2026.06 | 79.42 | 77.98 | — | — | |
| BiCBuffer=50/class2026.04 | 79.28 | 80.97 | — | — | |
| DyToxBuffer=50/class2026.04 | 79.21 | 81.23 | — | — | |
| DyToxModel=Vision Transformer (ViT-B/16)2026.06 | 79.21 | — | — | — | |
| RanPACBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 79.16 | 81.38 | — | — | |
| L2P*Backbone=CLIP, Shots=8-shot2026.06 | 79 | 78.06 | — | — | |
| L2PModel=Vision Transformer (ViT-B/16)2026.06 | 78.45 | — | — | — | |
| CP-PromptBackbone=CLIP, Shots=1-shot2026.06 | 78.28 | 77.23 | — | — | |
| CoReBackbone=ViT-B/16-IN21K2026.03 | 77.95 | 59.91 | — | — | |
| S-PromptBackbone=CLIP, Shots=1-shot2026.06 | 77.63 | 74.26 | — | — | |
| ViTBackbone=ViT-B/16-IN21K2026.03 | 77.55 | 61.49 | — | — | |
| AdapterBackbone=ViT-B/16-IN21K2026.03 | 76.5 | 58.53 | — | — | |
| SSFBackbone=ViT-B/16-IN21K2026.03 | 76.42 | 59.02 | — | — | |
| PromptBackbone=ViT-B/16-IN21K2026.03 | 76.2 | 57.2 | — | — | |
| FinetuneBackbone=ViT-B/16-IN21K2026.03 | 75.86 | 58.22 | — | — | |
| FinetuneBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 75.44 | 76.19 | — | — | |
| LwFModel=Vision Transformer (ViT-B/16)2026.06 | 75.35 | — | — | — | |
| EWCModel=Vision Transformer (ViT-B/16)2026.06 | 74.82 | — | — | — | |
| iCaRLBackbone=ViT-B/16-IN1K, Exemplars=10 per class2024.10 | 74.43 | 79.86 | — | — | |
| LwFScenario=Domain Incremental Learning2025.11 | 74.35 | — | 0.81 | — | |
| EWCScenario=Domain Incremental Learning2025.11 | 73.86 | — | 1.09 | — | |
| InfLORA*Backbone=CLIP, Shots=8-shot2026.06 | 73.12 | 66.1 | — | — | |
| SimpleCILBackbone=ViT-B/16-IN1K, Exemplars=02024.10 | 70.92 | 74.8 | — | — | |
| SimpleCILBuffer=0/class2026.04 | 70.92 | 74.8 | — | — | |
| LwF*Backbone=CLIP, Shots=8-shot2026.06 | 67.62 | 63.24 | — | — | |
| EwC*Backbone=CLIP, Shots=2-shot2026.06 | 66.47 | 58.11 | — | — | |
| InfLORA*Backbone=CLIP, Shots=2-shot2026.06 | 66.47 | 57.69 | — | — | |
| LwF*Backbone=CLIP, Shots=2-shot2026.06 | 66.03 | 57.48 | — | — | |
| EwC*Backbone=CLIP, Shots=8-shot2026.06 | 65.22 | 57.62 | — | — | |
| MEMOBackbone=ViT-B/16-IN1K, Exemplars=10 per class2024.10 | 64.8 | 68.24 | — | — | |
| LwF*Backbone=CLIP, Shots=4-shot2026.06 | 64.61 | 57.29 | — | — | |
| InfLORA*Backbone=CLIP, Shots=4-shot2026.06 | 63.9 | 58.27 | — | — | |
| EwC*Backbone=CLIP, Shots=4-shot2026.06 | 63.55 | 54.8 | — | — | |
| DualPrompt*Backbone=CLIP, Shots=8-shot2026.06 | 60.73 | 58.16 | — | — | |
| CODA-PromptBackbone=CLIP, Shots=8-shot2026.06 | 59.97 | 47.89 | — | — | |
| DualPrompt*Backbone=CLIP, Shots=2-shot2026.06 | 59.53 | 55.28 | — | — | |
| LwF*Backbone=CLIP, Shots=1-shot2026.06 | 59.4 | 52.99 | — | — | |
| EwC*Backbone=CLIP, Shots=1-shot2026.06 | 58.43 | 51.89 | — | — | |
| DualPrompt*Backbone=CLIP, Shots=4-shot2026.06 | 58.37 | 52.39 | — | — | |
| CODA-PromptBackbone=CLIP, Shots=4-shot2026.06 | 57.87 | 44.95 | — | — | |
| InfLORA*Backbone=CLIP, Shots=1-shot2026.06 | 57.44 | 50.36 | — | — |