Multi-Label Incremental Learning on MS COCO B0-C10 protocol official (val)
83.1mAP (Last)Joint
Evaluation results
| Method | Links | ||
|---|---|---|---|
| JointBackbone=ViT, PEFT=-, Buffer size=-2026.01 | 83.1 | — | |
| JointBackbone=CNN, PEFT=-, Buffer size=-2026.01 | 81.8 | — | |
| P2L-CA+Backbone=ViT, PEFT=✓, Buffer size=0/class2026.01 | 75.7 | 81.9 | |
| P2L-CABackbone=ViT, PEFT=✓, Buffer size=0/class2026.01 | 74.6 | 80.3 | |
| CSCBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 72.8 | 78 | |
| HCPBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 71.2 | 77.9 | |
| KRT-RBackbone=CNN, PEFT=×, Buffer size=20/class2026.01 | 70.2 | 76.5 | |
| CODABackbone=ViT, PEFT=✓, Buffer size=0/class2026.01 | 69.8 | 75.8 | |
| DualPromptBackbone=ViT, PEFT=✓, Buffer size=0/class2026.01 | 69.4 | 74.7 | |
| L2PBackbone=ViT, PEFT=✓, Buffer size=0/class2026.01 | 67.7 | 73.3 | |
| KRT*Backbone=ViT, PEFT=×, Buffer size=0/class2026.01 | 66.4 | 74.9 | |
| KRTBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 65.9 | 74.6 | |
| DER++Backbone=CNN, PEFT=×, Buffer size=20/class2026.01 | 63.1 | 72.7 | |
| AGCNBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 61.4 | 72.4 | |
| PODNetBackbone=CNN, PEFT=×, Buffer size=20/class2026.01 | 58.8 | 70 | |
| ERBackbone=CNN, PEFT=×, Buffer size=20/class2026.01 | 47.2 | 60.3 | |
| LWFBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 28.9 | 47.9 | |
| oEWCBackbone=CNN, PEFT=×, Buffer size=0/class2026.01 | 24.3 | 46.9 | |
| Fine-TuningBackbone=-, PEFT=-, Buffer size=-2026.01 | 16.9 | 38.3 |