Table 7.

The resulting precision, recall, and f1-score values for Scenario 4 (training the algorithms with the training dataset of Japan and testing on the Taiwan, China, Japan, and all test datasets separately).

		U-Net			ResU-Net
Size	32	64	128	32	64	128
Precision	44.13	32.01	48.56	42.16	61.03	41.14
Recall	56.46	66.64	56.97	57.31	47.01	63.58
F1-score	49.54	43.25	52.43	48.58	53.11	49.95
Precision	28.75	21.72	35.17	27.07	42.34	29.58
Recall	51.46	80.87	59.38	52.51	24.7	73.95
F1-score	36.89	34.24	44.17	35.73	31.2	42.26
Precision	63.52	63.46	67.96	67.47	71.82	62.95
Recall	80.18	78.3	74.18	73.55	74.88	73.55
F1-score	70.89	70.1	70.93	70.38	73.32	67.84
Precision	67.58	66.44	70.89	66.53	69.85	68.91
Recall	58.51	53.89	53.85	59.36	62.79	54.35
F1-score	62.72	59.51	61.21	62.74	66.13	60.77
Algorithms trained by sample from			All	Taiwan	China	Japan

The highest values of precision, recall, and f1-score are indicated in bold.