Class-Incremental Learning on ObjectNet benchmark leaderboard · SOTA2 Research

Class-Incremental Learning on ObjectNet

83.8Average AccuracyJoint full fine-tuning

Evaluation Results

Method	Links
Joint full fine-tuning2023.07		83.8	—
RanPAC2023.07		79.9	—
SGDS2026.01		76.97	66.51
TUNA2026.01		76.46	66.32
SAEF2026.02		76.38	66.58
DAF2026.01		76.11	66.37
MOS2026.01		74.69	63.62
MOS2026.01		74.69	63.62
MOS2026.02		74.69	63.62
SSIAT2026.01		73.65	62.45
SSIAT2026.01		73.65	62.45
SSIAT2026.02		73.65	62.45
RanPAC2026.01		72.76	62.02
RanPAC2026.01		72.76	62.02
RanPAC2026.02		72.76	62.02
TUNA2026.01		72.61	58.52
SLCA2026.01		72.55	61.3
SLCA2026.01		72.55	61.3
SLCA2026.02		72.55	61.3
EASE2026.01		70.84	57.86
EASE2026.01		70.84	57.86
EASE2026.02		70.84	57.86
APER + Adapter2026.01		67.18	55.24
APER + Adapter2026.01		67.18	55.24
APER + Adapter2026.02		67.18	55.24
CODA-Prompt2026.01		66.07	53.29
CODA-Prompt2026.01		66.07	53.29
CODA-Prompt2026.02		66.07	53.29
SimpleCIL2026.01		65.45	53.59
SimpleCIL2026.01		65.45	53.59
SimpleCIL2026.02		65.45	53.59
L2P2026.01		63.78	52.19
L2P2026.01		63.78	52.19
L2P2026.02		63.78	52.19
SPA2026.05		59.98	47.06
DualPrompt2026.01		59.27	49.33
DualPrompt2026.01		59.27	49.33
DualPrompt2026.02		59.27	49.33
BOFA2026.05		59.21	46.95
CLG-CBM2026.05		58.53	45.11
PROOF2026.05		56.07	43.69
SPA2026.05		53.28	47.59
DualPrompt2026.05		52.62	40.72
SimpleCIL2026.05		52.06	40.13
BOFA2026.05		51.89	46.76
L2P2026.05		51.4	39.39
CLG-CBM2026.05		49.8	43.56
DualPrompt2026.05		49.08	42.92
L2P2026.05		48.91	42.83
PROOF2026.05		48.9	43.62
RAPF2026.05		48.67	27.43
CODA-Prompt2026.05		46.49	34.13
SimpleCIL2026.05		45.11	40.13
CODA-Prompt2026.05		40.57	34.13
RAPF2026.05		39.28	28.73
ZS-CLIP2026.05		38.43	26.43
ZS-CLIP2026.05		31.12	26.43
Finetune2026.01		19.14	8.73
Finetune2026.01		19.14	8.73
Finetune2026.02		19.14	8.73