Mural classification model based on high- and low-level vision fusion

Heritage Science

Table 4 Classification results of various models

Category	Index	Mehdipour	P. Cheng	Lee	G. Cheng	Lin	TFNet
Buddha	Precision	59.62	50.00	59.62	53.70	61.54	72.22
	Recall rate	57.39	47.62	58.17	51.95	59.22	70.18
	F1-score	58.48	48.78	58.89	52.81	60.36	71.19
Bodhisattva	Precision	78.43	74.23	73.53	73.20	74.51	81.44
	Recall rate	77.10	72.98	73.02	71.49	72.16	80.04
	F1-score	77.76	73.60	73.27	72.33	73.32	80.73
Disciple	Precision	51.67	47.54	50.00	44.26	50.00	45.90
	Recall rate	49.09	46.56	49.12	42.10	48.55	45.01
	F1-score	50.35	47.04	49.56	43.15	49.26	45.45
Secular person	Precision	90.57	86.08	89.06	87.76	87.55	91.14
	Recall rate	88.72	85.43	87.09	85.48	84.99	89.57
	F1-score	89.64	85.75	88.06	86.60	86.25	90.35
Animal	Precision	64.52	67.74	67.74	56.99	66.67	70.97
	Recall rate	63.06	64.28	66.53	56.04	64.30	69.15
	F1-score	63.78	65.96	67.13	56.51	65.46	70.05
Plant	Precision	72.15	51.25	65.82	57.50	67.09	61.25
	Recall rate	70.18	50.25	63.89	57.01	66.44	60.98
	F1-score	71.15	50.75	64.84	57.25	66.76	61.11
Building	Precision	69.09	74.07	72.73	68.52	69.09	83.33
	Recall rate	68.22	72.55	71.02	67.77	68.05	82.83
	F1-score	68.65	73.30	71.86	68.14	68.57	83.08
Auspicious cloud	Precision	84.78	84.09	93.48	72.73	84.78	90.91
	Recall rate	82.51	82.11	91.45	71.93	84.00	90.04
	F1-score	83.63	83.09	92.45	72.33	84.39	90.47
Average	Precision	78.42	73.86	77.64	72.03	76.31	80.64
	Recall rate	76.60	71.25	75.80	69.86	74.73	78.06
	F1-score	77.05	71.91	76.28	70.39	75.08	78.63

The maximum value of precision rate, recall rate and F1-score for each class of all models are in italics