Continual Generalized Zero-Shot Learning on CUB benchmark leaderboard · SOTA2 Research

Continual Generalized Zero-Shot Learning on CUB

68.18Mean Accuracy (mSA)CA +mem

Evaluation Results

Method
CA +mem2022.03	68.18	42.44	50.68
NM-ZSL2022.03	64.91	46.05	53.79
CV +mem2022.03	63.16	27.5	37.84
Seq-CADA2022.03	55.55	26.96	35.62
NM-ZSL2022.03	55.45	43.25	47.04
unified feature-generative framework2022.03	50.51	52.2	51.18
A-CZSL2022.03	47.34	27.67	32.77
DVGR-CZSL2022.03	47.28	22.58	29.77
Tf-GCZSL2022.03	46.63	32.42	36.31
CV +mem2022.03	44.89	13.45	20.15
DVGR-CZSL2022.03	44.87	14.55	21.66
Tf-GCZSL2022.03	44.52	43.21	43.85
CA +mem2022.03	43.96	32.77	36.06
AGEM2022.03	43.4	—	—
unified feature-generative framework2022.03	42.39	37.09	36.35
EWC2022.03	40.95	—	—
Seq-CADA2022.03	40.82	14.37	21.14
AGEM2022.03	39.54	—	—
Seq-CVAE2022.03	38.95	20.89	27.19
EWC2022.03	35.89	—	—
A-CZSL2022.03	34.47	12	17.15
Seq-CVAE2022.03	24.66	8.57	12.18
Seq-fCLSWGAN2022.03	13.14	7.47	9.53
Seq-fCLSWGAN2022.03	11.16	6.54	8.25