TABLE 5.

Performance results of cross-database validation.

Models	Train OIQA/Test CVIQD			Train CVIQD/Test OIQA
	PLCC	SRCC	RMSE	PLCC	SRCC	RMSE
BMPRI (Min et al., 2018)	0.4904	0.2417	12.1862	0.7595	0.7205	1.3249
CEIQ (Yan et al., 2019)	0.6953	0.5470	9.9767	0.5012	0.4860	1.7856
BRISQUE (Mittal et al., 2012)	0.6166	0.5503	11.1772	0.4950	0.4054	1.8217
DIIVINE (Moorthy and Bovik, 2011)	0.5658	0.4114	11.4963	0.4454	0.3575	1.8904
SSEQ (Liu et al., 2014)	0.6175	0.6113	10.8955	0.4927	0.4568	1.7922
NRSL (Li et al., 2016)	0.6884	0.6199	10.4646	0.3651	0.2648	1.9431
OG-IQA (Liu et al., 2016)	0.6963	0.6392	10.1059	0.5154	0.5299	1.8076
SSP-BOIQA (Zheng et al., 2020)	0.726	0.705	9.588	0.627	0.601	–
MC360IQA (Sun et al., 2019)	0.8230	0.8140	7.8110	0.6816	0.5238	1.5471
Zhou Y. et al. (2022)	0.847	0.825	7.721	0.735	0.741	–
S3DAVS	0.8358	0.8125	7.9331	0.7817	0.6859	1.3938

The best-performing results are highlighted in bold.