Compositional Zero-Shot Learning on MIT-States open-vocabulary benchmark leaderboard · SOTA2 Research

Compositional Zero-Shot Learning on MIT-States open-vocabulary

28.9HMDFSP + SPA

Evaluation Results

Method	Links
DFSP + SPA2026.03		28.9	12.32	43.63	33.52	34.11	27.36	21.99	26.88	24.07
DFSP2026.03		28.12	12.09	44.19	33.13	35.81	25.11	18.08	26.51	21.71
SPA Avg2026.03		27.91	11.12	38.91	33.82	30.63	26.38	24.54	26.55	27.92
CSP + SPA2026.03		27.8	10.59	36.53	34.01	28.47	27.9	26.52	27.38	29.44
HPL + SPA2026.03		27.49	10.6	36.45	34.2	29.6	24.11	25.48	25.71	32.03
Troika + SPA2026.03		27.45	10.97	39.03	33.56	30.32	26.16	24.15	26.22	26.15
Baseline Avg2026.03		27.21	10.84	39.09	33.02	31.07	25.89	21.94	25.87	23.66
Troika2026.03		27.11	10.89	39.6	32.7	30.32	28.45	21.78	26.41	18.16
CSP2026.03		26.82	10.22	36.53	33.11	29.03	26.19	22.36	26.22	25.52
HPL2026.03		26.77	10.15	36.05	33.14	29.11	23.81	25.52	24.35	29.25
OADis+NEL2026.03		10.94	2.41	29.02	11.13	14.11	18.87	8.24	5.49	3.54
CANet2026.03		10.52	2.4	26.42	9.54	16.56	23.08	6.15	4.08	0.58
CompCos2026.03		10.22	1.97	26.53	10.29	14.32	21.09	5.86	2.89	0.63
SCEN2026.03		9.72	1.73	22.08	8.25	11.85	30.02	3.82	0.33	0.08
OADis2026.03		9.55	1.83	25.35	10.79	12.18	16.06	6.4	5.41	1.34
LE2026.03		7.64	1.01	16.29	9.46	10.24	11.38	5.98	4.15	2.87